x86/insn: Add some Intel instructions to the opcode map

Add to the opcode map the following instructions:
        cldemote
        tpause
        umonitor
        umwait
        movdiri
        movdir64b
        enqcmd
        enqcmds
        encls
        enclu
        enclv
        pconfig
        wbnoinvd

For information about the instructions, refer Intel SDM May 2019
(325462-070US) and Intel Architecture Instruction Set Extensions
May 2019 (319433-037).

The instruction decoding can be tested using the perf tools'
"x86 instruction decoder - new instructions" test as folllows:

  $ perf test -v "new " 2>&1 | grep -i cldemote
  Decoded ok: 0f 1c 00                    cldemote (%eax)
  Decoded ok: 0f 1c 05 78 56 34 12        cldemote 0x12345678
  Decoded ok: 0f 1c 84 c8 78 56 34 12     cldemote 0x12345678(%eax,%ecx,8)
  Decoded ok: 0f 1c 00                    cldemote (%rax)
  Decoded ok: 41 0f 1c 00                 cldemote (%r8)
  Decoded ok: 0f 1c 04 25 78 56 34 12     cldemote 0x12345678
  Decoded ok: 0f 1c 84 c8 78 56 34 12     cldemote 0x12345678(%rax,%rcx,8)
  Decoded ok: 41 0f 1c 84 c8 78 56 34 12  cldemote 0x12345678(%r8,%rcx,8)
  $ perf test -v "new " 2>&1 | grep -i tpause
  Decoded ok: 66 0f ae f3                 tpause %ebx
  Decoded ok: 66 0f ae f3                 tpause %ebx
  Decoded ok: 66 41 0f ae f0              tpause %r8d
  $ perf test -v "new " 2>&1 | grep -i umonitor
  Decoded ok: 67 f3 0f ae f0              umonitor %ax
  Decoded ok: f3 0f ae f0                 umonitor %eax
  Decoded ok: 67 f3 0f ae f0              umonitor %eax
  Decoded ok: f3 0f ae f0                 umonitor %rax
  Decoded ok: 67 f3 41 0f ae f0           umonitor %r8d
  $ perf test -v "new " 2>&1 | grep -i umwait
  Decoded ok: f2 0f ae f0                 umwait %eax
  Decoded ok: f2 0f ae f0                 umwait %eax
  Decoded ok: f2 41 0f ae f0              umwait %r8d
  $ perf test -v "new " 2>&1 | grep -i movdiri
  Decoded ok: 0f 38 f9 03                 movdiri %eax,(%ebx)
  Decoded ok: 0f 38 f9 88 78 56 34 12     movdiri %ecx,0x12345678(%eax)
  Decoded ok: 48 0f 38 f9 03              movdiri %rax,(%rbx)
  Decoded ok: 48 0f 38 f9 88 78 56 34 12  movdiri %rcx,0x12345678(%rax)
  $ perf test -v "new " 2>&1 | grep -i movdir64b
  Decoded ok: 66 0f 38 f8 18              movdir64b (%eax),%ebx
  Decoded ok: 66 0f 38 f8 88 78 56 34 12  movdir64b 0x12345678(%eax),%ecx
  Decoded ok: 67 66 0f 38 f8 1c           movdir64b (%si),%bx
  Decoded ok: 67 66 0f 38 f8 8c 34 12     movdir64b 0x1234(%si),%cx
  Decoded ok: 66 0f 38 f8 18              movdir64b (%rax),%rbx
  Decoded ok: 66 0f 38 f8 88 78 56 34 12  movdir64b 0x12345678(%rax),%rcx
  Decoded ok: 67 66 0f 38 f8 18           movdir64b (%eax),%ebx
  Decoded ok: 67 66 0f 38 f8 88 78 56 34 12       movdir64b 0x12345678(%eax),%ecx
  $ perf test -v "new " 2>&1 | grep -i enqcmd
  Decoded ok: f2 0f 38 f8 18              enqcmd (%eax),%ebx
  Decoded ok: f2 0f 38 f8 88 78 56 34 12  enqcmd 0x12345678(%eax),%ecx
  Decoded ok: 67 f2 0f 38 f8 1c           enqcmd (%si),%bx
  Decoded ok: 67 f2 0f 38 f8 8c 34 12     enqcmd 0x1234(%si),%cx
  Decoded ok: f3 0f 38 f8 18              enqcmds (%eax),%ebx
  Decoded ok: f3 0f 38 f8 88 78 56 34 12  enqcmds 0x12345678(%eax),%ecx
  Decoded ok: 67 f3 0f 38 f8 1c           enqcmds (%si),%bx
  Decoded ok: 67 f3 0f 38 f8 8c 34 12     enqcmds 0x1234(%si),%cx
  Decoded ok: f2 0f 38 f8 18              enqcmd (%rax),%rbx
  Decoded ok: f2 0f 38 f8 88 78 56 34 12  enqcmd 0x12345678(%rax),%rcx
  Decoded ok: 67 f2 0f 38 f8 18           enqcmd (%eax),%ebx
  Decoded ok: 67 f2 0f 38 f8 88 78 56 34 12       enqcmd 0x12345678(%eax),%ecx
  Decoded ok: f3 0f 38 f8 18              enqcmds (%rax),%rbx
  Decoded ok: f3 0f 38 f8 88 78 56 34 12  enqcmds 0x12345678(%rax),%rcx
  Decoded ok: 67 f3 0f 38 f8 18           enqcmds (%eax),%ebx
  Decoded ok: 67 f3 0f 38 f8 88 78 56 34 12       enqcmds 0x12345678(%eax),%ecx
  $ perf test -v "new " 2>&1 | grep -i enqcmds
  Decoded ok: f3 0f 38 f8 18              enqcmds (%eax),%ebx
  Decoded ok: f3 0f 38 f8 88 78 56 34 12  enqcmds 0x12345678(%eax),%ecx
  Decoded ok: 67 f3 0f 38 f8 1c           enqcmds (%si),%bx
  Decoded ok: 67 f3 0f 38 f8 8c 34 12     enqcmds 0x1234(%si),%cx
  Decoded ok: f3 0f 38 f8 18              enqcmds (%rax),%rbx
  Decoded ok: f3 0f 38 f8 88 78 56 34 12  enqcmds 0x12345678(%rax),%rcx
  Decoded ok: 67 f3 0f 38 f8 18           enqcmds (%eax),%ebx
  Decoded ok: 67 f3 0f 38 f8 88 78 56 34 12       enqcmds 0x12345678(%eax),%ecx
  $ perf test -v "new " 2>&1 | grep -i encls
  Decoded ok: 0f 01 cf                    encls
  Decoded ok: 0f 01 cf                    encls
  $ perf test -v "new " 2>&1 | grep -i enclu
  Decoded ok: 0f 01 d7                    enclu
  Decoded ok: 0f 01 d7                    enclu
  $ perf test -v "new " 2>&1 | grep -i enclv
  Decoded ok: 0f 01 c0                    enclv
  Decoded ok: 0f 01 c0                    enclv
  $ perf test -v "new " 2>&1 | grep -i pconfig
  Decoded ok: 0f 01 c5                    pconfig
  Decoded ok: 0f 01 c5                    pconfig
  $ perf test -v "new " 2>&1 | grep -i wbnoinvd
  Decoded ok: f3 0f 09                    wbnoinvd
  Decoded ok: f3 0f 09                    wbnoinvd

Signed-off-by: Adrian Hunter <adrian.hunter@intel.com>
Reviewed-by: Andi Kleen <ak@linux.intel.com>
Acked-by: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Borislav Petkov <bp@alien8.de>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: x86@kernel.org
Link: http://lore.kernel.org/lkml/20191115135447.6519-3-adrian.hunter@intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

x86/insn: perf tools: Add some instructions to the new instructions test

Add to the "x86 instruction decoder - new instructions" test the following
instructions:
cldemote
tpause
umonitor
umwait
movdiri
movdir64b
enqcmd
enqcmds
encls
enclu
enclv
pconfig
wbnoinvd

For information about the instructions, refer Intel SDM May 2019
(325462-070US) and Intel Architecture Instruction Set Extensions
May 2019 (319433-037).

Signed-off-by: Adrian Hunter <adrian.hunter@intel.com>
Reviewed-by: Andi Kleen <ak@linux.intel.com>
Cc: Borislav Petkov <bp@alien8.de>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: x86@kernel.org
Link: http://lore.kernel.org/lkml/20191115135447.6519-2-adrian.hunter@intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf map: Move seldom used ->flags field to second cacheline

So we start with:

  $ pahole -C map ~/bin/perf
  struct map {
   union {
   struct rb_node rb_node __attribute__((__aligned__(8))); /*     0    24 */
   struct list_head node;                   /*     0    16 */
   } __attribute__((__aligned__(8)));                                               /*     0    24 */
   u64                        start;                /*    24     8 */
   u64                        end;                  /*    32     8 */
   _Bool                      erange_warned:1;      /*    40: 0  1 */
   _Bool                      priv:1;               /*    40: 1  1 */

   /* XXX 6 bits hole, try to pack */
   /* XXX 3 bytes hole, try to pack */

   u32                        prot;                 /*    44     4 */
   u32                        flags;                /*    48     4 */

   /* XXX 4 bytes hole, try to pack */

   u64                        pgoff;                /*    56     8 */
   /* --- cacheline 1 boundary (64 bytes) --- */
   u64                        reloc;                /*    64     8 */
   u32                        maj;                  /*    72     4 */
   u32                        min;                  /*    76     4 */
   u64                        ino;                  /*    80     8 */
   u64                        ino_generation;       /*    88     8 */
   u64                        (*map_ip)(struct map *, u64); /*    96     8 */
   u64                        (*unmap_ip)(struct map *, u64); /*   104     8 */
   struct dso *               dso;                  /*   112     8 */
   refcount_t                 refcnt;               /*   120     4 */

   /* size: 128, cachelines: 2, members: 17 */
   /* sum members: 116, holes: 2, sum holes: 7 */
   /* sum bitfield members: 2 bits, bit holes: 1, sum bit holes: 6 bits */
   /* padding: 4 */
   /* forced alignments: 1 */
  } __attribute__((__aligned__(8)));
  $

and 'flags' is seldom used when printing details about the map or with
the "cacheline" sort order, we can move them it to the second cacheline,
that will allow combining it with 'refcnt', that is only four bytes:

  $ pahole -C map ~/bin/perf
  struct map {
   union {
   struct rb_node rb_node __attribute__((__aligned__(8))); /*     0    24 */
   struct list_head node;                   /*     0    16 */
   } __attribute__((__aligned__(8)));                                               /*     0    24 */
   u64                        start;                /*    24     8 */
   u64                        end;                  /*    32     8 */
   _Bool                      erange_warned:1;      /*    40: 0  1 */
   _Bool                      priv:1;               /*    40: 1  1 */

   /* XXX 6 bits hole, try to pack */
   /* XXX 3 bytes hole, try to pack */

   u32                        prot;                 /*    44     4 */
   u64                        pgoff;                /*    48     8 */
   u64                        reloc;                /*    56     8 */
   /* --- cacheline 1 boundary (64 bytes) --- */
   u32                        maj;                  /*    64     4 */
   u32                        min;                  /*    68     4 */
   u64                        ino;                  /*    72     8 */
   u64                        ino_generation;       /*    80     8 */
   u64                        (*map_ip)(struct map *, u64); /*    88     8 */
   u64                        (*unmap_ip)(struct map *, u64); /*    96     8 */
   struct dso *               dso;                  /*   104     8 */
   refcount_t                 refcnt;               /*   112     4 */
   u32                        flags;                /*   116     4 */

   /* size: 120, cachelines: 2, members: 17 */
   /* sum members: 116, holes: 1, sum holes: 3 */
   /* sum bitfield members: 2 bits, bit holes: 1, sum bit holes: 6 bits */
   /* forced alignments: 1 */
   /* last cacheline: 56 bytes */
  } __attribute__((__aligned__(8)));
  $

Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Link: https://lkml.kernel.org/n/tip-2cdw3zlw1mkamaf7nqtdlxfi@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf map: Use bitmap for booleans

The map->priv and map->erange_warned are seldom used, the first only in
tests/vmlinux-kallsyms.c, the later only when hist_entry__inc_addr_samples()
returns -ERANGE in 'perf top', which are really rare occasions, so make
them a bool bitfield.

This will open up space for other members on the first cacheline.

  $ pahole -C map ~/bin/perf
  struct map {
   union {
   struct rb_node rb_node __attribute__((__aligned__(8))); /*     0    24 */
   struct list_head node;                   /*     0    16 */
   } __attribute__((__aligned__(8)));                                               /*     0    24 */
   u64                        start;                /*    24     8 */
   u64                        end;                  /*    32     8 */
   _Bool                      erange_warned:1;      /*    40: 0  1 */
   _Bool                      priv:1;               /*    40: 1  1 */

   /* XXX 6 bits hole, try to pack */
   /* XXX 3 bytes hole, try to pack */

   u32                        prot;                 /*    44     4 */
   u32                        flags;                /*    48     4 */

   /* XXX 4 bytes hole, try to pack */

   u64                        pgoff;                /*    56     8 */
   /* --- cacheline 1 boundary (64 bytes) --- */
   u64                        reloc;                /*    64     8 */
   u32                        maj;                  /*    72     4 */
   u32                        min;                  /*    76     4 */
   u64                        ino;                  /*    80     8 */
   u64                        ino_generation;       /*    88     8 */
   u64                        (*map_ip)(struct map *, u64); /*    96     8 */
   u64                        (*unmap_ip)(struct map *, u64); /*   104     8 */
   struct dso *               dso;                  /*   112     8 */
   refcount_t                 refcnt;               /*   120     4 */

   /* size: 128, cachelines: 2, members: 17 */
   /* sum members: 116, holes: 2, sum holes: 7 */
   /* sum bitfield members: 2 bits, bit holes: 1, sum bit holes: 6 bits */
   /* padding: 4 */
   /* forced alignments: 1 */
  } __attribute__((__aligned__(8)));
  $

Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Link: https://lkml.kernel.org/n/tip-g5545pcq4ff0wr17tfb1piqt@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

libtraceevent: Fix parsing of event %o and %X argument types

Add missing "%o" and "%X". Ext4 events use "%o" for printing i_mode.

Signed-off-by: Konstantin Khlebnikov <khlebnikov@yandex-team.ru>
Reviewed-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
Cc: Tzvetomir Stoyanov (VMware) <tz.stoyanov@gmail.com>
Link: http://lore.kernel.org/lkml/157338066113.6548.11461421296091086041.stgit@buzz
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf callchain: Fix segfault in thread__resolve_callchain_sample()

Do not dereference 'chain' when it is NULL.

  $ perf record -e intel_pt//u -e branch-misses:u uname
  $ perf report --itrace=l --branch-history
  perf: Segmentation fault

Fixes: e9024d519d89 ("perf callchain: Honour the ordering of PERF_CONTEXT_{USER,KERNEL,etc}")
Signed-off-by: Adrian Hunter <adrian.hunter@intel.com>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Link: http://lore.kernel.org/lkml/20191114142538.4097-1-adrian.hunter@intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf map_groups: Auto sort maps by name, if needed

There are still lots of lookups by name, even if just when loading
vmlinux, till that code is studied to figure out if its possible to do
away with those map lookup by names, provide a way to sort it using
libc's qsort/bsearch.

Doing it at the first lookup defers the sorting a bit, and as the code
stands now, is never done for user maps, just for the kernel ones.

  # perf probe -l
  # perf probe -x ~/bin/perf -L __map_groups__find_by_name
  <__map_groups__find_by_name@/home/acme/git/perf/tools/perf/util/symbol.c:0>
        0  static struct map *__map_groups__find_by_name(struct map_groups *mg, const char *name)
        1  {
                  struct map **mapp;

        4         if (mg->maps_by_name == NULL &&
        5             map__groups__sort_by_name_from_rbtree(mg))
        6                 return NULL;

        8         mapp = bsearch(name, mg->maps_by_name, mg->nr_maps, sizeof(*mapp), map__strcmp_name);
        9         if (mapp)
       10                 return *mapp;
       11         return NULL;
       12  }

           struct map *map_groups__find_by_name(struct map_groups *mg, const char *name)
           {

  # perf probe -x ~/bin/perf 'found=__map_groups__find_by_name:10 name:string'
  Added new event:
    probe_perf:found     (on __map_groups__find_by_name:10 in /home/acme/bin/perf with name:string)

  You can now use it in all perf tools, such as:

   perf record -e probe_perf:found -aR sleep 1

  #
  # perf probe -x ~/bin/perf -L map_groups__find_by_name
  <map_groups__find_by_name@/home/acme/git/perf/tools/perf/util/symbol.c:0>
        0  struct map *map_groups__find_by_name(struct map_groups *mg, const char *name)
        1  {
        2         struct maps *maps = &mg->maps;
                  struct map *map;

        5         down_read(&maps->lock);

        7         if (mg->last_search_by_name && strcmp(mg->last_search_by_name->dso->short_name, name) == 0) {
        8                 map = mg->last_search_by_name;
        9                 goto out_unlock;
                  }
                  /*
                   * If we have mg->maps_by_name, then the name isn't in the rbtree,
                   * as mg->maps_by_name mirrors the rbtree when lookups by name are
                   * made.
                   */
       16         map = __map_groups__find_by_name(mg, name);
       17         if (map || mg->maps_by_name != NULL)
       18                 goto out_unlock;

                  /* Fallback to traversing the rbtree... */
       21         maps__for_each_entry(maps, map)
       22                 if (strcmp(map->dso->short_name, name) == 0) {
       23                         mg->last_search_by_name = map;
       24                         goto out_unlock;
                          }

       27         map = NULL;

           out_unlock:
       30         up_read(&maps->lock);
       31         return map;
       32  }

           int dso__load_vmlinux(struct dso *dso, struct map *map,
                                const char *vmlinux, bool vmlinux_allocated)

  # perf probe -x ~/bin/perf 'fallback=map_groups__find_by_name:21 name:string'
  Added new events:
    probe_perf:fallback  (on map_groups__find_by_name:21 in /home/acme/bin/perf with name:string)
    probe_perf:fallback_1 (on map_groups__find_by_name:21 in /home/acme/bin/perf with name:string)

  You can now use it in all perf tools, such as:

   perf record -e probe_perf:fallback_1 -aR sleep 1

  #
  # perf probe -l
    probe_perf:fallback  (on map_groups__find_by_name:21@util/symbol.c in /home/acme/bin/perf with name_string)
    probe_perf:fallback_1 (on map_groups__find_by_name:21@util/symbol.c in /home/acme/bin/perf with name_string)
    probe_perf:found     (on __map_groups__find_by_name:10@util/symbol.c in /home/acme/bin/perf with name_string)
  #
  # perf stat -e probe_perf:*

Now run 'perf top' in another term and then, after a while, stop 'perf stat':

Furthermore, if we ask for interval printing, we can see that that is done just
at the start of the workload:

  # perf stat -I1000 -e probe_perf:*
  #           time             counts unit events
       1.000319513                  0      probe_perf:found
       1.000319513                  0      probe_perf:fallback_1
       1.000319513                  0      probe_perf:fallback
       2.001868092             23,251      probe_perf:found
       2.001868092                  0      probe_perf:fallback_1
       2.001868092                  0      probe_perf:fallback
       3.002901597                  0      probe_perf:found
       3.002901597                  0      probe_perf:fallback_1
       3.002901597                  0      probe_perf:fallback
       4.003358591                  0      probe_perf:found
       4.003358591                  0      probe_perf:fallback_1
       4.003358591                  0      probe_perf:fallback
  ^C
  #

Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Link: https://lkml.kernel.org/n/tip-c5lmbyr14x448rcfii7y6t3k@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf machine: No need to check if kernel module maps pre-exist

We'only populating maps for kernel modules either from perf.data file
PERF_RECORD_MMAP records or when parsing /proc/modules, so there is no
need to first look if we already have those module maps in the list,
that would mean the kernel has duplicate entries.

So ditch one use of looking up maps by name.

Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Link: https://lkml.kernel.org/n/tip-gnzjg2hhuz6jnrw91m35059y@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf record: No need to process the synthesized MMAP events twice

At the end of a 'perf record' session, by default, we'll process all
samples and populate the threads, maps, etc so as to find out which of
the DSOs got samples, to reduce the size of the build-id table we'll
add to the perf.data headers.

But we don't need to process the PERF_RECORD_MMAP events synthesized
for the kernel modules, as we have those already via
perf_session__create_kernel_maps(), so add mmap/mmap2 handlers that
first look at event->header.misc to see if the event is for a user map,
bailing out if not.

Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Link: https://lkml.kernel.org/n/tip-mofoxvcx2dryppcw3o689jdd@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf map: No need to adjust the long name of modules

At some point in the past we needed to make sure we would get the long
name of modules and not just what we get from /proc/modules, but that
need, as described in the cset that introduced the adjustment function:

Fixes: c03d5184f0e9 ("perf machine: Adjust dso->long_name for offline module")
Without using the buildid-cache:

  # lsmod | grep trusted
  # insmod trusted.ko
  # lsmod | grep trusted
  trusted                24576  0
  # strace -e open,openat perf probe -m ./trusted.ko key_seal |& grep trusted
  openat(AT_FDCWD, "/sys/module/trusted/notes/.note.gnu.build-id", O_RDONLY) = 4
  openat(AT_FDCWD, "/sys/module/trusted/notes/.note.gnu.build-id", O_RDONLY) = 7
  openat(AT_FDCWD, "/root/trusted.ko", O_RDONLY) = 3
  openat(AT_FDCWD, "/root/.debug/root/trusted.ko/dd3d355d567394d540f527e093e0f64b95879584/probes", O_RDWR|O_CREAT, 0644) = 3
  openat(AT_FDCWD, "/usr/lib/debug/root/trusted.ko.debug", O_RDONLY) = -1 ENOENT (No such file or directory)
  openat(AT_FDCWD, "/usr/lib/debug/root/trusted.ko", O_RDONLY) = -1 ENOENT (No such file or directory)
  openat(AT_FDCWD, "/root/.debug/trusted.ko", O_RDONLY) = -1 ENOENT (No such file or directory)
  openat(AT_FDCWD, "/root/trusted.ko", O_RDONLY) = 3
  openat(AT_FDCWD, "trusted.ko.debug", O_RDONLY) = -1 ENOENT (No such file or directory)
  openat(AT_FDCWD, ".debug/trusted.ko.debug", O_RDONLY) = -1 ENOENT (No such file or directory)
  openat(AT_FDCWD, "trusted.ko.debug", O_RDONLY) = -1 ENOENT (No such file or directory)
  openat(AT_FDCWD, "/root/trusted.ko", O_RDONLY) = 3
  openat(AT_FDCWD, "/root/trusted.ko", O_RDONLY) = 3
  openat(AT_FDCWD, "/root/trusted.ko", O_RDONLY) = 4
  openat(AT_FDCWD, "/root/trusted.ko", O_RDONLY) = 3
    probe:key_seal       (on key_seal in trusted)
  # perf probe -l
    probe:key_seal       (on key_seal in trusted)
  #

No attempt at opening '[trusted]'.

Now using the build-id cache:

  # rmmod trusted
  # perf buildid-cache --add ./trusted.ko
  # insmod trusted.ko
  # strace -e open,openat perf probe -m ./trusted.ko key_seal |& grep trusted
  openat(AT_FDCWD, "/sys/module/trusted/notes/.note.gnu.build-id", O_RDONLY) = 4
  openat(AT_FDCWD, "/sys/module/trusted/notes/.note.gnu.build-id", O_RDONLY) = 7
  openat(AT_FDCWD, "/root/trusted.ko", O_RDONLY) = 3
  openat(AT_FDCWD, "/root/.debug/root/trusted.ko/dd3d355d567394d540f527e093e0f64b95879584/probes", O_RDWR|O_CREAT, 0644) = 3
  openat(AT_FDCWD, "/usr/lib/debug/root/trusted.ko.debug", O_RDONLY) = -1 ENOENT (No such file or directory)
  openat(AT_FDCWD, "/usr/lib/debug/root/trusted.ko", O_RDONLY) = -1 ENOENT (No such file or directory)
  openat(AT_FDCWD, "/root/.debug/trusted.ko", O_RDONLY) = -1 ENOENT (No such file or directory)
  openat(AT_FDCWD, "/root/trusted.ko", O_RDONLY) = 3
  openat(AT_FDCWD, "trusted.ko.debug", O_RDONLY) = -1 ENOENT (No such file or directory)
  openat(AT_FDCWD, ".debug/trusted.ko.debug", O_RDONLY) = -1 ENOENT (No such file or directory)
  openat(AT_FDCWD, "trusted.ko.debug", O_RDONLY) = -1 ENOENT (No such file or directory)
  openat(AT_FDCWD, "/root/trusted.ko", O_RDONLY) = 3
  openat(AT_FDCWD, "/root/trusted.ko", O_RDONLY) = 3
  openat(AT_FDCWD, "/root/trusted.ko", O_RDONLY) = 4
  openat(AT_FDCWD, "/root/trusted.ko", O_RDONLY) = 3
  #

Again, no attempt at reading '[trusted]'.

Finally, adding a probe to that function and then using:

[root@quaco ~]# perf trace -e probe_perf:*/max-stack=16/ --max-events=2
     0.000 perf/13456 probe_perf:dso__adjust_kmod_long_name(__probe_ip: 5492263)
                                       dso__adjust_kmod_long_name (/home/acme/bin/perf)
                                       machine__process_kernel_mmap_event (/home/acme/bin/perf)
                                       machine__process_mmap_event (/home/acme/bin/perf)
                                       perf_event__process_mmap (/home/acme/bin/perf)
                                       machines__deliver_event (/home/acme/bin/perf)
                                       perf_session__deliver_event (/home/acme/bin/perf)
                                       perf_session__process_event (/home/acme/bin/perf)
                                       process_simple (/home/acme/bin/perf)
                                       reader__process_events (/home/acme/bin/perf)
                                       __perf_session__process_events (/home/acme/bin/perf)
                                       perf_session__process_events (/home/acme/bin/perf)
                                       process_buildids (/home/acme/bin/perf)
                                       record__finish_output (/home/acme/bin/perf)
                                       __cmd_record (/home/acme/bin/perf)
                                       cmd_record (/home/acme/bin/perf)
                                       run_builtin (/home/acme/bin/perf)
     0.055 perf/13456 probe_perf:dso__adjust_kmod_long_name(__probe_ip: 5492263)
                                       dso__adjust_kmod_long_name (/home/acme/bin/perf)
                                       machine__process_kernel_mmap_event (/home/acme/bin/perf)
                                       machine__process_mmap_event (/home/acme/bin/perf)
                                       perf_event__process_mmap (/home/acme/bin/perf)
                                       machines__deliver_event (/home/acme/bin/perf)
                                       perf_session__deliver_event (/home/acme/bin/perf)
                                       perf_session__process_event (/home/acme/bin/perf)
                                       process_simple (/home/acme/bin/perf)
                                       reader__process_events (/home/acme/bin/perf)
                                       __perf_session__process_events (/home/acme/bin/perf)
                                       perf_session__process_events (/home/acme/bin/perf)
                                       process_buildids (/home/acme/bin/perf)
                                       record__finish_output (/home/acme/bin/perf)
                                       __cmd_record (/home/acme/bin/perf)
                                       cmd_record (/home/acme/bin/perf)
                                       run_builtin (/home/acme/bin/perf)
  #

This was the only path I could find using the perf tools that reach at this
function, then as of november/2019, if we put a probe in the line where the
actuall setting of the dso->long_name is done:

  # perf trace -e probe_perf:*
  ^C[root@quaco ~]
  # perf stat -e probe_perf:*  -I 2000
       2.000404265                  0      probe_perf:dso__adjust_kmod_long_name
       4.001142200                  0      probe_perf:dso__adjust_kmod_long_name
       6.001704120                  0      probe_perf:dso__adjust_kmod_long_name
       8.002398316                  0      probe_perf:dso__adjust_kmod_long_name
      10.002984010                  0      probe_perf:dso__adjust_kmod_long_name
      12.003597851                  0      probe_perf:dso__adjust_kmod_long_name
      14.004113303                  0      probe_perf:dso__adjust_kmod_long_name
      16.004582773                  0      probe_perf:dso__adjust_kmod_long_name
      18.005176373                  0      probe_perf:dso__adjust_kmod_long_name
      20.005801605                  0      probe_perf:dso__adjust_kmod_long_name
      22.006467540                  0      probe_perf:dso__adjust_kmod_long_name
  ^C    23.683261941                  0      probe_perf:dso__adjust_kmod_long_name

  #

Its not being used at all.

To further test this I used kvm.ko as the offline module, i.e. removed
if from the buildid-cache by nuking it completely (rm -rf ~/.debug) and
moved it from the normal kernel distro path, removed the modules, stoped
the kvm guest, and then installed it manually, etc.

  # rmmod kvm-intel
  # rmmod kvm
  # lsmod | grep kvm
  # modprobe kvm-intel
  modprobe: ERROR: ctx=0x55d3b1722260 path=/lib/modules/5.3.8-200.fc30.x86_64/kernel/arch/x86/kvm/kvm.ko.xz error=No such file or directory
  modprobe: ERROR: ctx=0x55d3b1722260 path=/lib/modules/5.3.8-200.fc30.x86_64/kernel/arch/x86/kvm/kvm.ko.xz error=No such file or directory
  modprobe: ERROR: could not insert 'kvm_intel': Unknown symbol in module, or unknown parameter (see dmesg)
  # insmod ./kvm.ko
  # modprobe kvm-intel
  modprobe: ERROR: ctx=0x562f34026260 path=/lib/modules/5.3.8-200.fc30.x86_64/kernel/arch/x86/kvm/kvm.ko.xz error=No such file or directory
  modprobe: ERROR: ctx=0x562f34026260 path=/lib/modules/5.3.8-200.fc30.x86_64/kernel/arch/x86/kvm/kvm.ko.xz error=No such file or directory
  # lsmod | grep kvm
  kvm_intel             299008  0
  kvm                   765952  1 kvm_intel
  irqbypass              16384  1 kvm
  #
  # perf probe -x ~/bin/perf machine__findnew_module_map:12 mname=m.name:string filename=filename:string 'dso_long_name=map->dso->long_name:string' 'dso_name=map->dso->name:string'
  # perf probe -l
    probe_perf:machine__findnew_module_map (on machine__findnew_module_map:12@util/machine.c in /home/acme/bin/perf with mname filename dso_long_name dso_name)
  # perf record
  ^C[ perf record: Woken up 2 times to write data ]
  [ perf record: Captured and wrote 3.416 MB perf.data (33956 samples) ]
  # perf trace -e probe_perf:machine*
  <SNIP>
       6.322 perf/23099 probe_perf:machine__findnew_module_map(__probe_ip: 5492493, mname: "[salsa20_generic]", filename: "/lib/modules/5.3.8-200.fc30.x86_64/kernel/crypto/salsa20_generic.ko.xz", dso_long_name: "/lib/modules/5.3.8-200.fc30.x86_64/kernel/crypto/salsa20_generic.ko.xz", dso_name: "[salsa20_generic]")
       6.375 perf/23099 probe_perf:machine__findnew_module_map(__probe_ip: 5492493, mname: "[kvm]", filename: "[kvm]", dso_long_name: "[kvm]", dso_name: "[kvm]")
  <SNIP>

The filename doesn't come with the path, no point in trying to set the dso->long_name.

  [root@quaco ~]# strace -e open,openat perf probe -m ./kvm.ko kvm_apic_local_deliver |& egrep 'open.*kvm'
  openat(AT_FDCWD, "/sys/module/kvm_intel/notes/.note.gnu.build-id", O_RDONLY) = 4
  openat(AT_FDCWD, "/sys/module/kvm/notes/.note.gnu.build-id", O_RDONLY) = 4
  openat(AT_FDCWD, "/lib/modules/5.3.8-200.fc30.x86_64/kernel/arch/x86/kvm", O_RDONLY|O_NONBLOCK|O_CLOEXEC|O_DIRECTORY) = 7
  openat(AT_FDCWD, "/sys/module/kvm_intel/notes/.note.gnu.build-id", O_RDONLY) = 8
  openat(AT_FDCWD, "/root/kvm.ko", O_RDONLY) = 3
  openat(AT_FDCWD, "/root/.debug/root/kvm.ko/5955f426cb93f03f30f3e876814be2db80ab0b55/probes", O_RDWR|O_CREAT, 0644) = 3
  openat(AT_FDCWD, "/usr/lib/debug/root/kvm.ko.debug", O_RDONLY) = -1 ENOENT (No such file or directory)
  openat(AT_FDCWD, "/usr/lib/debug/root/kvm.ko", O_RDONLY) = -1 ENOENT (No such file or directory)
  openat(AT_FDCWD, "/root/.debug/kvm.ko", O_RDONLY) = -1 ENOENT (No such file or directory)
  openat(AT_FDCWD, "/root/kvm.ko", O_RDONLY) = 3
  openat(AT_FDCWD, "kvm.ko.debug", O_RDONLY) = -1 ENOENT (No such file or directory)
  openat(AT_FDCWD, ".debug/kvm.ko.debug", O_RDONLY) = -1 ENOENT (No such file or directory)
  openat(AT_FDCWD, "kvm.ko.debug", O_RDONLY) = -1 ENOENT (No such file or directory)
  openat(AT_FDCWD, "/root/kvm.ko", O_RDONLY) = 3
  openat(AT_FDCWD, "/root/kvm.ko", O_RDONLY) = 3
  openat(AT_FDCWD, "/root/kvm.ko", O_RDONLY) = 4
  openat(AT_FDCWD, "/root/kvm.ko", O_RDONLY) = 3
  [root@quaco ~]#

Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Wang Nan <wangnan0@huawei.com>
Link: https://lkml.kernel.org/n/tip-jlfew3lyb24d58egrp0o72o2@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf map_groups: Add a front end cache for map lookups by name

Lets see if it helps:

First look at the probeable lines for the function that does lookups by
name in a map_groups struct:

  # perf probe -x ~/bin/perf -L map_groups__find_by_name
  <map_groups__find_by_name@/home/acme/git/perf/tools/perf/util/symbol.c:0>
        0  struct map *map_groups__find_by_name(struct map_groups *mg, const char *name)
        1  {
        2         struct maps *maps = &mg->maps;
                  struct map *map;

        5         down_read(&maps->lock);

        7         if (mg->last_search_by_name && strcmp(mg->last_search_by_name->dso->short_name, name) == 0) {
        8                 map = mg->last_search_by_name;
        9                 goto out_unlock;
                  }

       12         maps__for_each_entry(maps, map)
       13                 if (strcmp(map->dso->short_name, name) == 0) {
       14                         mg->last_search_by_name = map;
       15                         goto out_unlock;
                          }

       18         map = NULL;

           out_unlock:
       21         up_read(&maps->lock);
       22         return map;
       23  }

           int dso__load_vmlinux(struct dso *dso, struct map *map,
                                const char *vmlinux, bool vmlinux_allocated)

  #

Now add a probe to the place where we reuse the last search:

  # perf probe -x ~/bin/perf map_groups__find_by_name:8
  Added new event:
    probe_perf:map_groups__find_by_name (on map_groups__find_by_name:8 in /home/acme/bin/perf)

  You can now use it in all perf tools, such as:

   perf record -e probe_perf:map_groups__find_by_name -aR sleep 1

  #

Now lets do a system wide 'perf stat' counting those events:

  # perf stat -e probe_perf:*

Leave it running and lets do a 'perf top', then, after a while, stop the
'perf stat':

  # perf stat -e probe_perf:*
  ^C
   Performance counter stats for 'system wide':

               3,603      probe_perf:map_groups__find_by_name

        44.565253139 seconds time elapsed
  #

yeah, good to have.

Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Link: https://lkml.kernel.org/n/tip-tcz37g3nxv3tvxw3q90vga3p@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf maps: Do not use an rbtree to sort by map name

This is only used for the kernel maps, shave 24 bytes out 'struct map'
and just traverse the existing per ip rbtree to look for maps by name,
use a front end cache to reuse the last search if its the same name.

After this 'struct map' is down to just two cachelines:

  $ pahole -C map ~/bin/perf
  struct map {
   union {
   struct rb_node rb_node __attribute__((__aligned__(8))); /*     0    24 */
   struct list_head node;                   /*     0    16 */
   } __attribute__((__aligned__(8)));                                               /*     0    24 */
   u64                        start;                /*    24     8 */
   u64                        end;                  /*    32     8 */
   _Bool                      erange_warned;        /*    40     1 */

   /* XXX 3 bytes hole, try to pack */

   u32                        priv;                 /*    44     4 */
   u32                        prot;                 /*    48     4 */
   u32                        flags;                /*    52     4 */
   u64                        pgoff;                /*    56     8 */
   /* --- cacheline 1 boundary (64 bytes) --- */
   u64                        reloc;                /*    64     8 */
   u32                        maj;                  /*    72     4 */
   u32                        min;                  /*    76     4 */
   u64                        ino;                  /*    80     8 */
   u64                        ino_generation;       /*    88     8 */
   u64                        (*map_ip)(struct map *, u64); /*    96     8 */
   u64                        (*unmap_ip)(struct map *, u64); /*   104     8 */
   struct dso *               dso;                  /*   112     8 */
   refcount_t                 refcnt;               /*   120     4 */

   /* size: 128, cachelines: 2, members: 17 */
   /* sum members: 121, holes: 1, sum holes: 3 */
   /* padding: 4 */
   /* forced alignments: 1 */
  } __attribute__((__aligned__(8)));
  $

Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Link: https://lkml.kernel.org/n/tip-bvr8fqfgzxtgnhnwt5sssx5g@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf maps: Purge the entries from maps->names in __maps__purge()

No need to iterate via the ->names rbtree, as all the entries there
as in maps->entries as well, reuse __maps__purge() for that.

Doing it this way we can kill maps__for_each_entry_by_name(),
maps__for_each_entry_by_name_safe(), maps__{first,next}_by_name().

Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Link: https://lkml.kernel.org/n/tip-ps0nrio8pydyo23rr2s696ue@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf scripts python: exported-sql-viewer.py: Fix use of TRUE with SQLite

Prior to version 3.23 SQLite does not support TRUE or FALSE, so always
use 1 and 0 for SQLite.

Fixes: 26c11206f433 ("perf scripts python: exported-sql-viewer.py: Use new 'has_calls' column")
Signed-off-by: Adrian Hunter <adrian.hunter@intel.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: stable@vger.kernel.org # v5.3+
Link: http://lore.kernel.org/lkml/20191113120206.26957-1-adrian.hunter@intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf vendor events power9: Fix commas so PMU event files are valid JSON

No functional change.

Remove extra commas in the power9 JSON files so that the files
can be parsed and validated by other utilities such as Python
that fail to parse invalid JSON.

Before:

  $ diffstat -l -p1 /wb/1.patch | while read filename ; do echo $filename ; cat $filename | json_verify ; done
  tools/perf/pmu-events/arch/powerpc/power9/cache.json
  parse error: invalid object key (must be a string)
                                          [   {,     "EventCode": "0x300
                       (right here) ------^
  JSON is invalid
  tools/perf/pmu-events/arch/powerpc/power9/floating-point.json
  parse error: invalid object key (must be a string)
                                          [   {,     "EventCode": "0x141
                       (right here) ------^
  JSON is invalid
  tools/perf/pmu-events/arch/powerpc/power9/frontend.json
  parse error: invalid object key (must be a string)
                                          [   {,     "EventCode": "0x250
                       (right here) ------^
  JSON is invalid
  tools/perf/pmu-events/arch/powerpc/power9/marked.json
  parse error: invalid object key (must be a string)
                                          [   {,     "EventCode": "0x301
                       (right here) ------^
  JSON is invalid
  tools/perf/pmu-events/arch/powerpc/power9/memory.json
  parse error: invalid object key (must be a string)
                                          [   {,     "EventCode": "0x300
                       (right here) ------^
  JSON is invalid
  tools/perf/pmu-events/arch/powerpc/power9/other.json
  parse error: invalid object key (must be a string)
                                          [   {,     "EventCode": "0x308
                       (right here) ------^
  JSON is invalid
  tools/perf/pmu-events/arch/powerpc/power9/pipeline.json
  parse error: invalid object key (must be a string)
                                          [   {,     "EventCode": "0x4D0
                       (right here) ------^
  JSON is invalid
  tools/perf/pmu-events/arch/powerpc/power9/pmc.json
  parse error: invalid object key (must be a string)
                                          [   {,     "EventCode": "0x200
                       (right here) ------^
  JSON is invalid
  tools/perf/pmu-events/arch/powerpc/power9/translation.json
  parse error: invalid object key (must be a string)
                                          [   {,     "EventCode": "0x1E"
                       (right here) ------^
  JSON is invalid
  $

After:

  $ diffstat -l -p1 /wb/1.patch | while read filename ; do echo $filename ; cat $filename | json_verify ; done
  tools/perf/pmu-events/arch/powerpc/power9/cache.json
  JSON is valid
  tools/perf/pmu-events/arch/powerpc/power9/floating-point.json
  JSON is valid
  tools/perf/pmu-events/arch/powerpc/power9/frontend.json
  JSON is valid
  tools/perf/pmu-events/arch/powerpc/power9/marked.json
  JSON is valid
  tools/perf/pmu-events/arch/powerpc/power9/memory.json
  JSON is valid
  tools/perf/pmu-events/arch/powerpc/power9/other.json
  JSON is valid
  tools/perf/pmu-events/arch/powerpc/power9/pipeline.json
  JSON is valid
  tools/perf/pmu-events/arch/powerpc/power9/pmc.json
  JSON is valid
  tools/perf/pmu-events/arch/powerpc/power9/translation.json
  JSON is valid
  $

Signed-off-by: James Clark <james.clark@arm.com>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Kevin Mooney <kevin.mooney@arm.com>
Cc: Madhavan Srinivasan <maddy@linux.vnet.ibm.com>
Cc: Mamatha Inamdar <mamatha4@linux.vnet.ibm.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ravi Bangoria <ravi.bangoria@linux.ibm.com>
Cc: nd@arm.com
Link: http://lore.kernel.org/lkml/20191112160342.26470-3-james.clark@arm.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf vendor events power8: Fix commas so PMU event files are valid JSON

No functional change.

Remove extra commas in the power8 JSON files so that the files
can be parsed and validated by other utilities such as Python
that fail to parse invalid JSON.

Committer testing:

Before:

  $ diffstat -l -p1 /wb/1.patch | while read filename ; do echo $filename ; cat $filename | json_verify ; done
  tools/perf/pmu-events/arch/powerpc/power8/cache.json
  parse error: invalid object key (must be a string)
                                          [   {,     "EventCode": "0x4c0
                       (right here) ------^
  JSON is invalid
  tools/perf/pmu-events/arch/powerpc/power8/floating-point.json
  parse error: invalid object key (must be a string)
                                          [   {,     "EventCode": "0x200
                       (right here) ------^
  JSON is invalid
  tools/perf/pmu-events/arch/powerpc/power8/frontend.json
  parse error: invalid object key (must be a string)
                                          [   {,     "EventCode": "0x250
                       (right here) ------^
  JSON is invalid
  tools/perf/pmu-events/arch/powerpc/power8/marked.json
  parse error: invalid object key (must be a string)
                                          [   {,     "EventCode": "0x351
                       (right here) ------^
  JSON is invalid
  tools/perf/pmu-events/arch/powerpc/power8/memory.json
  parse error: invalid object key (must be a string)
                                          [   {,     "EventCode": "0x100
                       (right here) ------^
  JSON is invalid
  tools/perf/pmu-events/arch/powerpc/power8/other.json
  parse error: invalid object key (must be a string)
                                          [   {,     "EventCode": "0x1f0
                       (right here) ------^
  JSON is invalid
  tools/perf/pmu-events/arch/powerpc/power8/pipeline.json
  parse error: invalid object key (must be a string)
                                          [   {,     "EventCode": "0x100
                       (right here) ------^
  JSON is invalid
  tools/perf/pmu-events/arch/powerpc/power8/pmc.json
  parse error: invalid object key (must be a string)
                                          [   {,     "EventCode": "0x200
                       (right here) ------^
  JSON is invalid
  tools/perf/pmu-events/arch/powerpc/power8/translation.json
  parse error: invalid object key (must be a string)
                                          [   {,     "EventCode": "0x4c0
                       (right here) ------^
  JSON is invalid
  $

After:

  $ diffstat -l -p1 /wb/1.patch | while read filename ; do echo $filename ; cat $filename | json_verify ; done
  tools/perf/pmu-events/arch/powerpc/power8/cache.json
  JSON is valid
  tools/perf/pmu-events/arch/powerpc/power8/floating-point.json
  JSON is valid
  tools/perf/pmu-events/arch/powerpc/power8/frontend.json
  JSON is valid
  tools/perf/pmu-events/arch/powerpc/power8/marked.json
  JSON is valid
  tools/perf/pmu-events/arch/powerpc/power8/memory.json
  JSON is valid
  tools/perf/pmu-events/arch/powerpc/power8/other.json
  JSON is valid
  tools/perf/pmu-events/arch/powerpc/power8/pipeline.json
  JSON is valid
  tools/perf/pmu-events/arch/powerpc/power8/pmc.json
  JSON is valid
  tools/perf/pmu-events/arch/powerpc/power8/translation.json
  JSON is valid
  $

Signed-off-by: James Clark <james.clark@arm.com>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Kevin Mooney <kevin.mooney@arm.com>
Cc: Madhavan Srinivasan <maddy@linux.vnet.ibm.com>
Cc: Mamatha Inamdar <mamatha4@linux.vnet.ibm.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ravi Bangoria <ravi.bangoria@linux.ibm.com>
Cc: nd@arm.com
Link: http://lore.kernel.org/lkml/20191112160342.26470-2-james.clark@arm.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf vendor events arm64: Fix commas so PMU event files are valid JSON

No functional change.

Add and remove extra commas in the arm64 JSON files so that the files
can be parsed and validated by other utilities such as Python that fail
to parse invalid JSON.

Committer testing:

Before:

  $ diffstat -l -p1 /wb/1.patch | while read filename ; do echo $filename ; cat $filename | json_verify ; done
  tools/perf/pmu-events/arch/arm64/ampere/emag/branch.json
  parse error: invalid object key (must be a string)
                                          [     {         "ArchStdEvent"
                       (right here) ------^
  JSON is invalid
  tools/perf/pmu-events/arch/arm64/ampere/emag/bus.json
  parse error: invalid object key (must be a string)
                                          [     {         "ArchStdEvent"
                       (right here) ------^
  JSON is invalid
  tools/perf/pmu-events/arch/arm64/ampere/emag/cache.json
  parse error: invalid object key (must be a string)
                                          [     {         "ArchStdEvent"
                       (right here) ------^
  JSON is invalid
  tools/perf/pmu-events/arch/arm64/ampere/emag/clock.json
  parse error: unallowed token at this point in JSON text
                                          [     {         "PublicDescrip
                       (right here) ------^
  JSON is invalid
  tools/perf/pmu-events/arch/arm64/ampere/emag/exception.json
  parse error: invalid object key (must be a string)
                                          [     {         "ArchStdEvent"
                       (right here) ------^
  JSON is invalid
  tools/perf/pmu-events/arch/arm64/ampere/emag/instruction.json
  parse error: invalid object key (must be a string)
                                          [     {         "ArchStdEvent"
                       (right here) ------^
  JSON is invalid
  tools/perf/pmu-events/arch/arm64/ampere/emag/intrinsic.json
  parse error: invalid object key (must be a string)
                                          [     {         "ArchStdEvent"
                       (right here) ------^
  JSON is invalid
  tools/perf/pmu-events/arch/arm64/ampere/emag/memory.json
  parse error: invalid object key (must be a string)
                                          [     {         "ArchStdEvent"
                       (right here) ------^
  JSON is invalid
  tools/perf/pmu-events/arch/arm64/ampere/emag/pipeline.json
  parse error: unallowed token at this point in JSON text
                                          [     {         "PublicDescrip
                       (right here) ------^
  JSON is invalid
  tools/perf/pmu-events/arch/arm64/arm/cortex-a53/branch.json
  parse error: invalid object key (must be a string)
                                          [   {     "ArchStdEvent":  "BR
                       (right here) ------^
  JSON is invalid
  tools/perf/pmu-events/arch/arm64/arm/cortex-a53/bus.json
  parse error: invalid object key (must be a string)
                                          [   {         "ArchStdEvent":
                       (right here) ------^
  JSON is invalid
  tools/perf/pmu-events/arch/arm64/arm/cortex-a53/other.json
  parse error: invalid object key (must be a string)
                                          [   {         "ArchStdEvent":
                       (right here) ------^
  JSON is invalid
  tools/perf/pmu-events/arch/arm64/arm/cortex-a57-a72/core-imp-def.json
  parse error: invalid object key (must be a string)
                                          [     {         "ArchStdEvent"
                       (right here) ------^
  JSON is invalid
  tools/perf/pmu-events/arch/arm64/armv8-recommended.json
  parse error: after array element, I expect ',' or ']'
                                          [     {         "PublicDescrip
                       (right here) ------^
  JSON is invalid
  tools/perf/pmu-events/arch/arm64/cavium/thunderx2/core-imp-def.json
  parse error: invalid object key (must be a string)
                                          [     {         "ArchStdEvent"
                       (right here) ------^
  JSON is invalid
  tools/perf/pmu-events/arch/arm64/hisilicon/hip08/core-imp-def.json
  parse error: invalid object key (must be a string)
                                          [     {         "ArchStdEvent"
                       (right here) ------^
  JSON is invalid
  tools/perf/pmu-events/arch/arm64/hisilicon/hip08/uncore-ddrc.json
  parse error: invalid object key (must be a string)
                                          [    {      "EventCode": "0x00
                       (right here) ------^
  JSON is invalid
  tools/perf/pmu-events/arch/arm64/hisilicon/hip08/uncore-hha.json
  parse error: invalid object key (must be a string)
                                          [    {      "EventCode": "0x00
                       (right here) ------^
  JSON is invalid
  tools/perf/pmu-events/arch/arm64/hisilicon/hip08/uncore-l3c.json
  parse error: invalid object key (must be a string)
                                          [    {      "EventCode": "0x00
                       (right here) ------^
  JSON is invalid
  $

After:

  $ diffstat -l -p1 /wb/1.patch | while read filename ; do echo $filename ; cat $filename | json_verify ; done
  tools/perf/pmu-events/arch/arm64/ampere/emag/branch.json
  JSON is valid
  tools/perf/pmu-events/arch/arm64/ampere/emag/bus.json
  JSON is valid
  tools/perf/pmu-events/arch/arm64/ampere/emag/cache.json
  JSON is valid
  tools/perf/pmu-events/arch/arm64/ampere/emag/clock.json
  JSON is valid
  tools/perf/pmu-events/arch/arm64/ampere/emag/exception.json
  JSON is valid
  tools/perf/pmu-events/arch/arm64/ampere/emag/instruction.json
  JSON is valid
  tools/perf/pmu-events/arch/arm64/ampere/emag/intrinsic.json
  JSON is valid
  tools/perf/pmu-events/arch/arm64/ampere/emag/memory.json
  JSON is valid
  tools/perf/pmu-events/arch/arm64/ampere/emag/pipeline.json
  JSON is valid
  tools/perf/pmu-events/arch/arm64/arm/cortex-a53/branch.json
  JSON is valid
  tools/perf/pmu-events/arch/arm64/arm/cortex-a53/bus.json
  JSON is valid
  tools/perf/pmu-events/arch/arm64/arm/cortex-a53/other.json
  JSON is valid
  tools/perf/pmu-events/arch/arm64/arm/cortex-a57-a72/core-imp-def.json
  JSON is valid
  tools/perf/pmu-events/arch/arm64/armv8-recommended.json
  JSON is valid
  tools/perf/pmu-events/arch/arm64/cavium/thunderx2/core-imp-def.json
  JSON is valid
  tools/perf/pmu-events/arch/arm64/hisilicon/hip08/core-imp-def.json
  JSON is valid
  tools/perf/pmu-events/arch/arm64/hisilicon/hip08/uncore-ddrc.json
  JSON is valid
  tools/perf/pmu-events/arch/arm64/hisilicon/hip08/uncore-hha.json
  JSON is valid
  tools/perf/pmu-events/arch/arm64/hisilicon/hip08/uncore-l3c.json
  JSON is valid
  $

Signed-off-by: James Clark <james.clark@arm.com>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: John Garry <john.garry@huawei.com>
Cc: Kevin Mooney <kevin.mooney@arm.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: nd@arm.com
Link: http://lore.kernel.org/lkml/20191112160342.26470-1-james.clark@arm.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf parse: Use YYABORT to clear stack after failure, plugging leaks

Using return rather than YYABORT means that the stack isn't cleared up
following a failure. The change to YYABORT means the return value is 1
rather than -1, but the callers just check for a result of 0 (success).
Add missing free of a list when an error occurs in event_pmu.

Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Link: http://lore.kernel.org/lkml/20191109075840.181231-1-irogers@google.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf tool: Provide an option to print perf_event_open args and return value

Perf record with verbose=2 already prints this information along with
whole lot of other traces which requires lot of scrolling. Introduce
an option to print only perf_event_open() arguments and return value.

Sample o/p:

  $ perf --debug perf-event-open=1 record -- ls > /dev/null
  ------------------------------------------------------------
  perf_event_attr:
    size                             112
    { sample_period, sample_freq }   4000
    sample_type                      IP|TID|TIME|PERIOD
    read_format                      ID
    disabled                         1
    inherit                          1
    exclude_kernel                   1
    mmap                             1
    comm                             1
    freq                             1
    enable_on_exec                   1
    task                             1
    precise_ip                       3
    sample_id_all                    1
    exclude_guest                    1
    mmap2                            1
    comm_exec                        1
    ksymbol                          1
    bpf_event                        1
  ------------------------------------------------------------
  sys_perf_event_open: pid 4308  cpu 0  group_fd -1  flags 0x8 = 4
  sys_perf_event_open: pid 4308  cpu 1  group_fd -1  flags 0x8 = 5
  sys_perf_event_open: pid 4308  cpu 2  group_fd -1  flags 0x8 = 6
  sys_perf_event_open: pid 4308  cpu 3  group_fd -1  flags 0x8 = 8
  sys_perf_event_open: pid 4308  cpu 4  group_fd -1  flags 0x8 = 9
  sys_perf_event_open: pid 4308  cpu 5  group_fd -1  flags 0x8 = 10
  sys_perf_event_open: pid 4308  cpu 6  group_fd -1  flags 0x8 = 11
  sys_perf_event_open: pid 4308  cpu 7  group_fd -1  flags 0x8 = 12
  ------------------------------------------------------------
  perf_event_attr:
    type                             1
    size                             112
    config                           0x9
    watermark                        1
    sample_id_all                    1
    bpf_event                        1
    { wakeup_events, wakeup_watermark } 1
  ------------------------------------------------------------
  sys_perf_event_open: pid -1  cpu 0  group_fd -1  flags 0x8
  sys_perf_event_open failed, error -13
  [ perf record: Woken up 1 times to write data ]
  [ perf record: Captured and wrote 0.002 MB perf.data (9 samples) ]

Committer notes:

Just like the 'verbose' variable this new 'debug_peo_args' needs to be
added to util/python.c, since we don't link the debug.o file in the
python binding, which ended up making 'perf test python' fail with:

  # perf test -v python
  18: 'import perf' in python                               :
  --- start ---
  test child forked, pid 19237
  Traceback (most recent call last):
    File "<stdin>", line 1, in <module>
  ImportError: /tmp/build/perf/python/perf.so: undefined symbol: debug_peo_args
  test child finished with -1
  ---- end ----
  'import perf' in python: FAILED!
  #

After adding that new variable to util/python.c:

  # perf test -v python
  18: 'import perf' in python                               :
  --- start ---
  test child forked, pid 22364
  test child finished with 0
  ---- end ----
  'import perf' in python: Ok
  #

Signed-off-by: Ravi Bangoria <ravi.bangoria@linux.ibm.com>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Link: http://lore.kernel.org/lkml/20191108094128.28769-1-ravi.bangoria@linux.ibm.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf map: Remove ->groups from 'struct map'

With this 'struct map' uses a bit over 3 cachelines:

  $ pahole -C map ~/bin/perf
  <SNIP>
   /* --- cacheline 2 boundary (128 bytes) --- */
   u64                        (*unmap_ip)(struct map *, u64); /*   128     8 */
   struct dso *               dso;                            /*   136     8 */
   refcount_t                 refcnt;                         /*   144     4 */

   /* size: 152, cachelines: 3, members: 18 */
   /* sum members: 145, holes: 1, sum holes: 3 */
   /* padding: 4 */
   /* forced alignments: 2 */
   /* last cacheline: 24 bytes */
  } __attribute__((__aligned__(8)));
  $

We probably can move map->map/unmap_ip() moved to 'struct map_groups',
that will shave more 16 bytes, getting this almost to two cachelines.

Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Link: https://lkml.kernel.org/n/tip-ymlv3nzpofv2fugnjnizkrwy@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf map: Combine maps__fixup_overlappings with its only use

In the process we can kill some of the struct map->groups usage, trying
to get rid of this per-full struct map fields getting in the way of
sharing a map across father/parent processes.

Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Link: https://lkml.kernel.org/n/tip-e50eqtqw3za24vmbjnqmmcs6@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf annotate: Stop using map->groups, use map_symbol->mg instead

These were the last uses of map->groups, next cset will nuke it.

Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Link: https://lkml.kernel.org/n/tip-n3g0foos7l7uxq9nar0zo0vj@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf tools: Add a 'struct map_groups' pointer to 'struct map_symbol'

And fill it whenever we setup a a 'struct map_symbol', now we need to
use it, next cset.

Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Link: https://lkml.kernel.org/n/tip-fzwfcnddenz1o7uj1fzw3g46@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf symbols: Use kmaps(map)->machine when we know its a kernel map

And then stop using map->groups to achieve that.

To test that that branch is being taken, probe the function that is only
called from there and then run something like 'perf top' in another
xterm:

  # perf probe -x ~/bin/perf machine__map_x86_64_entry_trampolines
  Added new event:
    probe_perf:machine__map_x86_64_entry_trampolines (on machine__map_x86_64_entry_trampolines in /home/acme/bin/perf)

  You can now use it in all perf tools, such as:

   perf record -e probe_perf:machine__map_x86_64_entry_trampolines -aR sleep 1

  # perf trace -e probe_perf:*
       0.000 bash/10614 probe_perf:machine__map_x86_64_entry_trampolines(__probe_ip: 5224944)
  ^C#

Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Link: https://lkml.kernel.org/n/tip-lgrrzdxo2p9liq2keivcg887@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

pref tools: Make 'struct addr_map_symbol' contain 'struct map_symbol'

So that we pass that substructure around and with it consolidate lots of
functions that receive a (map, symbol) pair and now can receive just a
'struct map_symbol' pointer.

This further paves the way to add 'struct map_groups' to 'struct
map_symbol' so that we can have all we need for annotation so that we
can ditch 'struct map'->groups, i.e. have the map_groups pointer in a
more central place, avoiding the pointer in the 'struct map' that have
tons of instances.

Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Link: https://lkml.kernel.org/n/tip-fs90ttd9q12l7989fo7pw81q@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf callchain: Use 'struct map_symbol' in 'struct callchain_cursor_node'

To ease passing around map+symbol, just like done for other parts of the
tree recently.

Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf unwind: Use 'struct map_symbol' in 'struct unwind_entry'

To help in passing that info around to callchain routines that, for the
same reason, are moving to use 'struct map_symbol'.

Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Link: https://lkml.kernel.org/n/tip-epsiibeprpxa8qpwji47uskc@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf annotate: Pass a 'map_symbol' in places receiving a pair of 'map' and 'symbol' pointers

We are already passing things like:

  symbol__annotate(ms->sym, ms->map, ...)

So shorten the signature of such functions to receive the 'map_symbol'
pointer.

This also paves the way to having the 'struct map_groups' pointer in the
'struct map_symbol' so that we can get rid of 'struct map'->groups.

Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Link: https://lkml.kernel.org/n/tip-23yx8v1t41nzpkpi7rdrozww@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf tools: Add map_groups to 'struct addr_location'

From there we can get al->mg->machine, so replace that field with the
more useful 'struct map_groups' that for now we're obtaining from
al->map->groups, and that is one thing getting into the way of maps
being fully shareable.

Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Link: https://lkml.kernel.org/n/tip-4qdducrm32tgrjupcp0kjh1e@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf map_groups: Pass the object to map_groups__find_ams()

We were just passing a map to look for and reuse its map->groups member,
but the idea is that this is going away, as a map can be in multiple
rb_trees when being reused via a map_node, so do as all the other
map_groups methods and pass as its first arg the object being operated
on.

Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Link: https://lkml.kernel.org/n/tip-nmi2pbggqloogwl6vxrvex5a@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf symbols: Stop using map->groups, we can use kmaps instead

To test that that function is being called I just added a probe on that
place, enabled it via 'perf trace' asking for at most 16 levels of
backtraces, system wide, and then ran 'perf top' on another xterm,
voilà:

  # perf probe -x ~/bin/perf dso__process_kernel_symbol
  Added new event:
    probe_perf:dso__process_kernel_symbol (on dso__process_kernel_symbol in /home/acme/bin/perf)

  You can now use it in all perf tools, such as:

   perf record -e probe_perf:dso__process_kernel_symbol -aR sleep 1

  # perf trace -e probe_perf:dso__process_kernel_symbol/max-stack=16/ --max-events=2
  # perf trace -e probe_perf:dso__process_kernel_symbol/max-stack=16/ --max-events=2
       0.000 :17345/17345 probe_perf:dso__process_kernel_symbol(__probe_ip: 5680224)
                                         dso__process_kernel_symbol (/home/acme/bin/perf)
                                         dso__load_vmlinux (/home/acme/bin/perf)
                                         dso__load_vmlinux_path (/home/acme/bin/perf)
                                         dso__load (/home/acme/bin/perf)
                                         map__load (/home/acme/bin/perf)
                                         thread__find_map (/home/acme/bin/perf)
                                         machine__resolve (/home/acme/bin/perf)
                                         deliver_event (/home/acme/bin/perf)
                                         __ordered_events__flush.part.0 (/home/acme/bin/perf)
                                         process_thread (/home/acme/bin/perf)
                                         start_thread (/usr/lib64/libpthread-2.29.so)
       0.064 :17345/17345 probe_perf:dso__process_kernel_symbol(__probe_ip: 5680224)
                                         dso__process_kernel_symbol (/home/acme/bin/perf)
                                         dso__load_vmlinux (/home/acme/bin/perf)
                                         dso__load_vmlinux_path (/home/acme/bin/perf)
                                         dso__load (/home/acme/bin/perf)
                                         map__load (/home/acme/bin/perf)
                                         thread__find_map (/home/acme/bin/perf)
                                         machine__resolve (/home/acme/bin/perf)
                                         deliver_event (/home/acme/bin/perf)
                                         __ordered_events__flush.part.0 (/home/acme/bin/perf)
                                         process_thread (/home/acme/bin/perf)
                                         start_thread (/usr/lib64/libpthread-2.29.so)
  #
  # perf stat -e probe_perf:dso__process_kernel_symbol
  ^C
   Performance counter stats for 'system wide':

           107,308      probe_perf:dso__process_kernel_symbol

       8.215399813 seconds time elapsed
  #

Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Link: https://lkml.kernel.org/n/tip-5fy66x5hr5ct9pmw84jkiwvm@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf map: Use map->dso->kernel + map__kmaps() in map__kmaps()

Its equivalent to using map->groups to obtain the machine struct.

Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Link: https://lkml.kernel.org/n/tip-bdbazuj4ggrmzxdviaqdrdwh@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

Merge tag 'perf-core-for-mingo-5.5-20191107' of git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux into perf/core

Pull perf/core improvements and fixes from Arnaldo Carvalho de Melo:

perf report:

  Jin Yao:

  - Introduce --total-cycles, for basic block profiling, further using data
    obtained from LBR, an example should suffice:

      # perf record -b
      ^C[ perf record: Woken up 595 times to write data ]
      [ perf record: Captured and wrote 156.672 MB perf.data (196873 samples) ]

      # perf evlist -v
      cycles: size: 112, { sample_period, sample_freq }: 4000, sample_type: IP|TID|TIME|CPU|PERIOD|BRANCH_STACK, read_format: ID, disabled: 1, inherit: 1, mmap: 1, comm: 1, freq: 1, task: 1, precise_ip: 3, sample_id_all: 1, exclude_guest: 1, mmap2: 1, comm_exec: 1, ksymbol: 1, bpf_event: 1, branch_sample_type: ANY

      # perf report --total-cycles --stdio
      # To display the perf.data header info, please use --header/--header-only options.
      #
      # Total Lost Samples: 0
      #
      # Samples: 6M of event 'cycles'
      # Event count (approx.): 6299936
      #
      # Sampled  Sampled   Avg     Avg
      # Cycles%  Cycles  Cycles%  Cycles                 [Program Block Range]     Shared Object
      # .......  ......  .......  .....   ....................................  ................
      #
         2.17%     1.7M   0.08%     607       [compiler.h:199 -> common.c:221]  [kernel.vmlinux]
         0.72%   544.5K   0.03%     230     [entry_64.S:657 -> entry_64.S:662]  [kernel.vmlinux]
         0.56%   541.8K   0.09%     672       [compiler.h:199 -> common.c:300]  [kernel.vmlinux]
         0.39%   293.2K   0.01%     104   [list_debug.c:43 -> list_debug.c:61]  [kernel.vmlinux]
         0.36%   278.6K   0.03%     272   [entry_64.S:1289 -> entry_64.S:1308]  [kernel.vmlinux]

perf record:

  Adrian Hunter:

  - Allow storing perf.data in a directory together with a copy of /proc/kcore.

  Jiwei Sun:

  - Add support for limit perf output file size, i.e.:

    # perf record --all-cpus -F 10000 --max-size=4M sleep 10h
    [ perf record: perf size limit reached (4097 KB), stopping session ]
    [ perf record: Woken up 6 times to write data ]
    [ perf record: Captured and wrote 4.048 MB perf.data (54094 samples) ]
    Terminated
    # ls -lah perf.data
    -rw-------. 1 root root 4.1M Nov  7 15:27 perf.data
    #

perf stat:

  Jiri Olsa:

  - Add --per-node agregation support:

    In live mode:

      # perf stat  -a -I 1000 -e cycles --per-node
      #           time node   cpus             counts unit events
           1.000542550 N0       20          6,202,097      cycles
           1.000542550 N1       20            639,559      cycles
           2.002040063 N0       20          7,412,495      cycles
           2.002040063 N1       20          2,185,577      cycles
           3.003451699 N0       20          6,508,917      cycles
           3.003451699 N1       20            765,607      cycles
      ...

    Or in the record/report stat session:

      # perf stat record -a -I 1000 -e cycles
      #           time             counts unit events
           1.000536937         10,008,468      cycles
           2.002090152          9,578,539      cycles
           3.003625233          7,647,869      cycles
           4.005135036          7,032,086      cycles
      ^C     4.340902364          3,923,893      cycles

      # perf stat report --per-node
      #           time node   cpus             counts unit events
           1.000536937 N0       20          9,355,086      cycles
           1.000536937 N1       20            653,382      cycles
           2.002090152 N0       20          7,712,838      cycles
           2.002090152 N1       20          1,865,701      cycles
       ...

perf probe:

  Masami Hiramatsu:

  Various fixes related to recent additions to the DWARF format:

  - Fix to find range-only function instance

  - Walk function lines in lexical blocks

  - Fix to show function entry line as probe-able

  - Fix wrong address verification

  - Fix to probe a function which has no entry pc

  - Fix to probe an inline function which has no entry pc

  - Fix to list probe event with correct line number

  - Fix to show inlined function callsite without entry_pc

  - Fix to show ranges of variables in functions without entry_pc

  - Return a better scope DIE if there is no best scope

  - Skip end-of-sequence and non statement lines

  - Filter out instances except for inlined subroutine and subprogram

  - Fix to show calling lines of inlined functions

  - Skip overlapped location on searching variables

perf inject:

  Adrian Hunter:

  - Do not strip evsels with --strip, as they are needed for create_gcov
    (see the autofdo example in tools/perf/Documentation/intel-pt.txt).

Intel PT:

  Adrian Hunter:

  - Intel PT uses an auxtrace_cache to store the results of code-walking, to avoid
    repeated decoding. Add an auxtrace_cache__remove to handle text poke events.

core:

  Andi Kleen:

  - Always preserve errno while cleaning up perf_event_open failures.

llvm:

  Arnaldo Carvalho de Melo:

  - No need to tell that the request for saving a .o file for BPF events, as
    expressed in ~/.perfconfig was satisfied, make that a debug message.

perf vendor events:

Intel:

  Haiyan Song:

  - Update CascadelakeX events to v1.05.

  - Update all the Intel JSON metrics from TMAM 3.6.

Treewide:

  Ian Rogers:

  - Improve error paths, plugging leaks found using LLVM tools
    such as libFuzzer.

jevents:

  Yunfeng Ye:

  - Fix resource leak in process_mapfile() and main()

perf kvm:

  Igor Lubashev:

  - Use evlist layer api when possible.

libsubcmd:

  James Clark:

  - Move EXTRA_FLAGS to the end to allow overriding existing flags.

  - Use -O0 with DEBUG=1

perf diff:

  Jin Yao:

  - Don't use hack to skip column length calculation

CoreSight ETM:

  Leo yan:

  - Fix definition of macro TO_CS_QUEUE_NR

ARM64:

  John Garry:

  - Do not try to include libelf header files when its feature detection
    failed, fixing the cross build for ARM64.

perf tests:

  Leo Yan:

  - Fix out of bounds memory access in the backward ring buffer test.

Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: Ingo Molnar <mingo@kernel.org>

perf/x86/amd: Remove set but not used variable 'active'

'-Wunused-but-set-variable' triggers this warning:

  arch/x86/events/amd/core.c: In function amd_pmu_handle_irq:
  arch/x86/events/amd/core.c:656:6: warning: variable active set but not used [-Wunused-but-set-variable]

GCC is right, 'active' is not used anymore.

This variable was introduced earlier this year and then removed in:

  df4d29732fdad perf/x86/amd: Change/fix NMI latency mitigation to use a timestamp

[ mingo: Improved the changelog, fixed build warning caused by this fix, improved surrounding code. ]

Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: Zheng Yongjun <zhengyongjun3@huawei.com>
Cc: <acme@kernel.org>
Cc: <alexander.shishkin@linux.intel.com>
Cc: <mark.rutland@arm.com>
Cc: <peterz@infradead.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Link: https://lkml.kernel.org/r/20191110094453.113001-1-zhengyongjun3@huawei.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>

Merge tag 'v5.4-rc7' into perf/core, to pick up fixes

Signed-off-by: Ingo Molnar <mingo@kernel.org>

Linux 5.4-rc7

Merge tag 'armsoc-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/soc/soc

Pull ARM SoC fixes from Olof Johansson:
 "A set of fixes that have trickled in over the last couple of weeks:

   - MAINTAINER update for Cavium/Marvell ThunderX2

   - stm32 tweaks to pinmux for Joystick/Camera, and RAM allocation for
     CAN interfaces

   - i.MX fixes for voltage regulator GPIO mappings, fixes voltage
     scaling issues

   - More i.MX fixes for various issues on i.MX eval boards: interrupt
     storm due to u-boot leaving pins in new states, fixing power button
     config, a couple of compatible-string corrections.

   - Powerdown and Suspend/Resume fixes for Allwinner A83-based tablets

   - A few documentation tweaks and a fix of a memory leak in the reset
     subsystem"

* tag 'armsoc-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/soc/soc:
  MAINTAINERS: update Cavium ThunderX2 maintainers
  ARM: dts: stm32: change joystick pinctrl definition on stm32mp157c-ev1
  ARM: dts: stm32: remove OV5640 pinctrl definition on stm32mp157c-ev1
  ARM: dts: stm32: Fix CAN RAM mapping on stm32mp157c
  ARM: dts: stm32: relax qspi pins slew-rate for stm32mp157
  arm64: dts: zii-ultra: fix ARM regulator GPIO handle
  ARM: sunxi: Fix CPU powerdown on A83T
  ARM: dts: sun8i-a83t-tbs-a711: Fix WiFi resume from suspend
  arm64: dts: imx8mn: fix compatible string for sdma
  arm64: dts: imx8mm: fix compatible string for sdma
  reset: fix reset_control_ops kerneldoc comment
  ARM: dts: imx6-logicpd: Re-enable SNVS power key
  soc: imx: gpc: fix initialiser format
  ARM: dts: imx6qdl-sabreauto: Fix storm of accelerometer interrupts
  arm64: dts: ls1028a: fix a compatible issue
  reset: fix reset_control_get_exclusive kerneldoc comment
  reset: fix reset_control_lookup kerneldoc comment
  reset: fix of_reset_control_get_count kerneldoc comment
  reset: fix of_reset_simple_xlate kerneldoc comment
  reset: Fix memory leak in reset_control_array_put()

Merge tag 'staging-5.4-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/staging

Pull IIO fixes and staging driver from Greg KH:
 "Here is a mix of a number of IIO driver fixes for 5.4-rc7, and a whole
  new staging driver.

  The IIO fixes resolve some reported issues, all are tiny.

  The staging driver addition is the vboxsf filesystem, which is the
  VirtualBox guest shared folder code. Hans has been trying to get
  filesystem reviewers to review the code for many months now, and
  Christoph finally said to just merge it in staging now as it is
  stand-alone and the filesystem people can review it easier over time
  that way.

  I know it's late for this big of an addition, but it is stand-alone.

  The code has been in linux-next for a while, long enough to pick up a
  few tiny fixes for it already so people are looking at it.

  All of these have been in linux-next with no reported issues"

* tag 'staging-5.4-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/staging:
  staging: Fix error return code in vboxsf_fill_super()
  staging: vboxsf: fix dereference of pointer dentry before it is null checked
  staging: vboxsf: Remove unused including <linux/version.h>
  staging: Add VirtualBox guest shared folder (vboxsf) support
  iio: adc: stm32-adc: fix stopping dma
  iio: imu: inv_mpu6050: fix no data on MPU6050
  iio: srf04: fix wrong limitation in distance measuring
  iio: imu: adis16480: make sure provided frequency is positive

Merge tag 'char-misc-5.4-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/char-misc

Pull char/misc driver fixes from Greg KH:
 "Here are a number of late-arrival driver fixes for issues reported for
  some char/misc drivers for 5.4-rc7

  These all come from the different subsystem/driver maintainers as
  things that they had reports for and wanted to see fixed.

  All of these have been in linux-next with no reported issues"

* tag 'char-misc-5.4-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/char-misc:
  intel_th: pci: Add Jasper Lake PCH support
  intel_th: pci: Add Comet Lake PCH support
  intel_th: msu: Fix possible memory leak in mode_store()
  intel_th: msu: Fix overflow in shift of an unsigned int
  intel_th: msu: Fix missing allocation failure check on a kstrndup
  intel_th: msu: Fix an uninitialized mutex
  intel_th: gth: Fix the window switching sequence
  soundwire: slave: fix scanf format
  soundwire: intel: fix intel_register_dai PDI offsets and numbers
  interconnect: Add locking in icc_set_tag()
  interconnect: qcom: Fix icc_onecell_data allocation
  soundwire: depend on ACPI || OF
  soundwire: depend on ACPI
  thunderbolt: Drop unnecessary read when writing LC command in Ice Lake
  thunderbolt: Fix lockdep circular locking depedency warning
  thunderbolt: Read DP IN adapter first two dwords in one go

Merge tag 'configfs-for-5.4-2' of git://git.infradead.org/users/hch/configfs

Pull configfs regression fix from Christoph Hellwig:
 "Fix a regression from this merge window in the configfs symlink
  handling (Honggang Li)"

* tag 'configfs-for-5.4-2' of git://git.infradead.org/users/hch/configfs:
  configfs: calculate the depth of parent item

Merge branch 'x86-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip

Pull x86 fixes from Thomas Gleixner:
 "A small set of fixes for x86:

   - Make the tsc=reliable/nowatchdog command line parameter work again.
     It was broken with the introduction of the early TSC clocksource.

   - Prevent the evaluation of exception stacks before they are set up.
     This causes a crash in dumpstack because the stack walk termination
     gets screwed up.

   - Prevent a NULL pointer dereference in the rescource control file
     system.

   - Avoid bogus warnings about APIC id mismatch related to the LDR
     which can happen when the LDR is not in use and therefore not
     initialized. Only evaluate that when the APIC is in logical
     destination mode"

* 'x86-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86/tsc: Respect tsc command line paraemeter for clocksource_tsc_early
  x86/dumpstack/64: Don't evaluate exception stacks before setup
  x86/apic/32: Avoid bogus LDR warnings
  x86/resctrl: Prevent NULL pointer dereference when reading mondata

Merge branch 'timers-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip

Pull timer fixes from Thomas Gleixner:
 "A small set of fixes for timekeepoing and clocksource drivers:

   - VDSO data was updated conditional on the availability of a VDSO
     capable clocksource. This causes the VDSO functions which do not
     depend on a VDSO capable clocksource to operate on stale data.
     Always update unconditionally.

   - Prevent a double free in the mediatek driver

   - Use the proper helper in the sh_mtu2 driver so it won't attempt to
     initialize non-existing interrupts"

* 'timers-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  timekeeping/vsyscall: Update VDSO data unconditionally
  clocksource/drivers/sh_mtu2: Do not loop using platform_get_irq_by_name()
  clocksource/drivers/mediatek: Fix error handling

Merge branch 'sched-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip

Pull scheduler fixes from Thomas Gleixner:
 "Two fixes for scheduler regressions:

   - Plug a subtle race condition which was introduced with the rework
     of the next task selection functionality. The change of task
     properties became unprotected which can be observed inconsistently
     causing state corruption.

   - A trivial compile fix for CONFIG_CGROUPS=n"

* 'sched-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  sched: Fix pick_next_task() vs 'change' pattern race
  sched/core: Fix compilation error when cgroup not selected

Merge branch 'perf-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip

Pull perf tooling fixes from Thomas Gleixner:

 - Fix the time sorting algorithm which was broken due to truncation of
   big numbers

 - Fix the python script generator fail caused by a broken tracepoint
   array iterator

* 'perf-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  perf tools: Fix time sorting
  perf tools: Remove unused trace_find_next_event()
  perf scripting engines: Iterate on tep event arrays directly

Merge branch 'irq-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip

Pull irq fixlet from Thomas Gleixner:
 "A trivial fix for a kernel doc regression where an argument change was
  not reflected in the documentation"

* 'irq-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  irq/irqdomain: Update __irq_domain_alloc_fwnode() function documentation

Merge branch 'core-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip

Pull stacktrace fix from Thomas Gleixner:
 "A small fix for a stacktrace regression.

  Saving a stacktrace for a foreign task skipped an extra entry which
  makes e.g. the output of /proc/$PID/stack incomplete"

* 'core-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  stacktrace: Don't skip first entry on noncurrent tasks

Merge tag '5.4-rc7-smb3-fix' of git://git.samba.org/sfrench/cifs-2.6

Pull cifs fix from Steve French:
 "Small fix for an smb3 reconnect bug (also marked for stable)"

* tag '5.4-rc7-smb3-fix' of git://git.samba.org/sfrench/cifs-2.6:
  SMB3: Fix persistent handles reconnect

lib: Remove select of inexistant GENERIC_IO

config option GENERIC_IO was removed but still selected by lib/kconfig
This patch finish the cleaning.

Fixes: 9de8da47742b ("kconfig: kill off GENERIC_IO option")
Acked-by: Rob Herring <robh@kernel.org>
Signed-off-by: Corentin Labbe <clabbe@baylibre.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Merge tag 'pinctrl-v5.4-3' of git://git.kernel.org/pub/scm/linux/kernel/git/linusw/linux-pinctrl

Pull pin control fixes from Linus Walleij:

 - Fix glitch risks in the Intel GPIO

 - Fix the Intel Cherryview valid irq mask calculation.

 - Allocate the Intel Cherryview irqchip dynamically.

 - Fix the valid mask init sequency on the ST STMFX driver.

* tag 'pinctrl-v5.4-3' of git://git.kernel.org/pub/scm/linux/kernel/git/linusw/linux-pinctrl:
  pinctrl: stmfx: fix valid_mask init sequence
  pinctrl: cherryview: Allocate IRQ chip dynamic
  pinctrl: cherryview: Fix irq_valid_mask calculation
  pinctrl: intel: Avoid potential glitches if pin is in GPIO mode

Merge tag 'for-5.4-rc6-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/kdave/linux

Pull btrfs fixes from David Sterba:
 "A few regressions and fixes for stable.

  Regressions:

   - fix a race leading to metadata space leak after task received a
     signal

   - un-deprecate 2 ioctls, marked as deprecated by mistake

  Fixes:

   - fix limit check for number of devices during chunk allocation

   - fix a race due to double evaluation of i_size_read inside max()
     macro, can cause a crash

   - remove wrong device id check in tree-checker"

* tag 'for-5.4-rc6-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/kdave/linux:
  btrfs: un-deprecate ioctls START_SYNC and WAIT_SYNC
  btrfs: save i_size to avoid double evaluation of i_size_read in compress_file_range
  Btrfs: fix race leading to metadata space leak after task received signal
  btrfs: tree-checker: Fix wrong check on max devid
  btrfs: Consider system chunk array size for new SYSTEM chunks

Merge tag 'linux-watchdog-5.4-rc7' of git://www.linux-watchdog.org/linux-watchdog

Pull watchdog fixes from Wim Van Sebroeck:

 - cpwd: fix build regression

 - pm8916_wdt: fix pretimeout registration flow

 - meson: Fix the wrong value of left time

 - imx_sc_wdt: Pretimeout should follow SCU firmware format

 - bd70528: Add MODULE_ALIAS to allow module auto loading

* tag 'linux-watchdog-5.4-rc7' of git://www.linux-watchdog.org/linux-watchdog:
  watchdog: bd70528: Add MODULE_ALIAS to allow module auto loading
  watchdog: imx_sc_wdt: Pretimeout should follow SCU firmware format
  watchdog: meson: Fix the wrong value of left time
  watchdog: pm8916_wdt: fix pretimeout registration flow
  watchdog: cpwd: fix build regression

Merge git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net

Pull networking fixes from David Miller:

 1) BPF sample build fixes from Björn Töpel

 2) Fix powerpc bpf tail call implementation, from Eric Dumazet.

 3) DCCP leaks jiffies on the wire, fix also from Eric Dumazet.

 4) Fix crash in ebtables when using dnat target, from Florian Westphal.

 5) Fix port disable handling whne removing bcm_sf2 driver, from Florian
    Fainelli.

 6) Fix kTLS sk_msg trim on fallback to copy mode, from Jakub Kicinski.

 7) Various KCSAN fixes all over the networking, from Eric Dumazet.

 8) Memory leaks in mlx5 driver, from Alex Vesker.

 9) SMC interface refcounting fix, from Ursula Braun.

10) TSO descriptor handling fixes in stmmac driver, from Jose Abreu.

11) Add a TX lock to synchonize the kTLS TX path properly with crypto
    operations. From Jakub Kicinski.

12) Sock refcount during shutdown fix in vsock/virtio code, from Stefano
    Garzarella.

13) Infinite loop in Intel ice driver, from Colin Ian King.

* git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net: (108 commits)
  ixgbe: need_wakeup flag might not be set for Tx
  i40e: need_wakeup flag might not be set for Tx
  igb/igc: use ktime accessors for skb->tstamp
  i40e: Fix for ethtool -m issue on X722 NIC
  iavf: initialize ITRN registers with correct values
  ice: fix potential infinite loop because loop counter being too small
  qede: fix NULL pointer deref in __qede_remove()
  net: fix data-race in neigh_event_send()
  vsock/virtio: fix sock refcnt holding during the shutdown
  net: ethernet: octeon_mgmt: Account for second possible VLAN header
  mac80211: fix station inactive_time shortly after boot
  net/fq_impl: Switch to kvmalloc() for memory allocation
  mac80211: fix ieee80211_txq_setup_flows() failure path
  ipv4: Fix table id reference in fib_sync_down_addr
  ipv6: fixes rt6_probe() and fib6_nh->last_probe init
  net: hns: Fix the stray netpoll locks causing deadlock in NAPI path
  net: usb: qmi_wwan: add support for DW5821e with eSIM support
  CDC-NCM: handle incomplete transfer of MTU
  nfc: netlink: fix double device reference drop
  NFC: st21nfca: fix double free
  ...

Merge tag 'for-linus-2019-11-08' of git://git.kernel.dk/linux-block

Pull block fixes from Jens Axboe:

 - Two NVMe device removal crash fixes, and a compat fixup for for an
   ioctl that was introduced in this release (Anton, Charles, Max - via
   Keith)

 - Missing error path mutex unlock for drbd (Dan)

 - cgroup writeback fixup on dead memcg (Tejun)

 - blkcg online stats print fix (Tejun)

* tag 'for-linus-2019-11-08' of git://git.kernel.dk/linux-block:
  cgroup,writeback: don't switch wbs immediately on dead wbs if the memcg is dead
  block: drbd: remove a stray unlock in __drbd_send_protocol()
  blkcg: make blkcg_print_stat() print stats only for online blkgs
  nvme: change nvme_passthru_cmd64 to explicitly mark rsvd
  nvme-multipath: fix crash in nvme_mpath_clear_ctrl_paths
  nvme-rdma: fix a segmentation fault during module unload

Merge branch '40GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/jkirsher/net-queue

Jeff Kirsher says:

====================
Intel Wired LAN Driver Fixes 2019-11-08

This series contains fixes to igb, igc, ixgbe, i40e, iavf and ice
drivers.

Colin Ian King fixes a potentially wrap-around counter in a for-loop.

Nick fixes the default ITR values for the iavf driver to 50 usecs
interval.

Arkadiusz fixes 'ethtool -m' for X722 devices where the correct value
cannot be obtained from the firmware, so add X722 to the check to ensure
the wrong value is not returned.

Jake fixes igb and igc drivers in their implementation of launch time
support by declaring skb->tstamp value as ktime_t instead of s64.

Magnus fixes ixgbe and i40e where the need_wakeup flag for transmit may
not be set for AF_XDP sockets that are only used to send packets.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>

ixgbe: need_wakeup flag might not be set for Tx

The need_wakeup flag for Tx might not be set for AF_XDP sockets that
are only used to send packets. This happens if there is at least one
outstanding packet that has not been completed by the hardware and we
get that corresponding completion (which will not generate an
interrupt since interrupts are disabled in the napi poll loop) between
the time we stopped processing the Tx completions and interrupts are
enabled again. In this case, the need_wakeup flag will have been
cleared at the end of the Tx completion processing as we believe we
will get an interrupt from the outstanding completion at a later point
in time. But if this completion interrupt occurs before interrupts
are enable, we lose it and should at that point really have set the
need_wakeup flag since there are no more outstanding completions that
can generate an interrupt to continue the processing. When this
happens, user space will see a Tx queue need_wakeup of 0 and skip
issuing a syscall, which means will never get into the Tx processing
again and we have a deadlock.

This patch introduces a quick fix for this issue by just setting the
need_wakeup flag for Tx to 1 all the time. I am working on a proper
fix for this that will toggle the flag appropriately, but it is more
challenging than I anticipated and I am afraid that this patch will
not be completed before the merge window closes, therefore this easier
fix for now. This fix has a negative performance impact in the range
of 0% to 4%. Towards the higher end of the scale if you have driver
and application on the same core and issue a lot of packets, and
towards no negative impact if you use two cores, lower transmission
speeds and/or a workload that also receives packets.

Signed-off-by: Magnus Karlsson <magnus.karlsson@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

i40e: need_wakeup flag might not be set for Tx

The need_wakeup flag for Tx might not be set for AF_XDP sockets that
are only used to send packets. This happens if there is at least one
outstanding packet that has not been completed by the hardware and we
get that corresponding completion (which will not generate an
interrupt since interrupts are disabled in the napi poll loop) between
the time we stopped processing the Tx completions and interrupts are
enabled again. In this case, the need_wakeup flag will have been
cleared at the end of the Tx completion processing as we believe we
will get an interrupt from the outstanding completion at a later point
in time. But if this completion interrupt occurs before interrupts
are enable, we lose it and should at that point really have set the
need_wakeup flag since there are no more outstanding completions that
can generate an interrupt to continue the processing. When this
happens, user space will see a Tx queue need_wakeup of 0 and skip
issuing a syscall, which means will never get into the Tx processing
again and we have a deadlock.

This patch introduces a quick fix for this issue by just setting the
need_wakeup flag for Tx to 1 all the time. I am working on a proper
fix for this that will toggle the flag appropriately, but it is more
challenging than I anticipated and I am afraid that this patch will
not be completed before the merge window closes, therefore this easier
fix for now. This fix has a negative performance impact in the range
of 0% to 4%. Towards the higher end of the scale if you have driver
and application on the same core and issue a lot of packets, and
towards no negative impact if you use two cores, lower transmission
speeds and/or a workload that also receives packets.

Signed-off-by: Magnus Karlsson <magnus.karlsson@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

igb/igc: use ktime accessors for skb->tstamp

When implementing launch time support in the igb and igc drivers, the
skb->tstamp value is assumed to be a s64, but it's declared as a ktime_t
value.

Although ktime_t is typedef'd to s64 it wasn't always, and the kernel
provides accessors for ktime_t values.

Use the ktime_to_timespec64 and ktime_set accessors instead of directly
assuming that the variable is always an s64.

This improves portability if the code is ever moved to another kernel
version, or if the definition of ktime_t ever changes again in the
future.

Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Acked-by: Vinicius Costa Gomes <vinicius.gomes@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

i40e: Fix for ethtool -m issue on X722 NIC

This patch contains fix for a problem with command:
'ethtool -m <dev>'
which breaks functionality of:
'ethtool <dev>'
when called on X722 NIC

Disallowed update of link phy_types on X722 NIC
Currently correct value cannot be obtained from FW
Previously wrong value returned by FW was used and was
a root cause for incorrect output of 'ethtool <dev>' command

Signed-off-by: Arkadiusz Kubalewski <arkadiusz.kubalewski@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

iavf: initialize ITRN registers with correct values

Since commit 92418fb14750 ("i40e/i40evf: Use usec value instead of reg
value for ITR defines") the driver tracks the interrupt throttling
intervals in single usec units, although the actual ITRN registers are
programmed in 2 usec units. Most register programming flows in the driver
correctly handle the conversion, although it is currently not applied when
the registers are initialized to their default values. Most of the time
this doesn't present a problem since the default values are usually
immediately overwritten through the standard adaptive throttling mechanism,
or updated manually by the user, but if adaptive throttling is disabled and
the interval values are left alone then the incorrect value will persist.

Since the intended default interval of 50 usecs (vs. 100 usecs as
programmed) performs better for most traffic workloads, this can lead to
performance regressions.

This patch adds the correct conversion when writing the initial values to
the ITRN registers.

Signed-off-by: Nicholas Nunley <nicholas.d.nunley@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

ice: fix potential infinite loop because loop counter being too small

Currently the for-loop counter i is a u8 however it is being checked
against a maximum value hw->num_tx_sched_layers which is a u16. Hence
there is a potential wrap-around of counter i back to zero if
hw->num_tx_sched_layers is greater than 255.  Fix this by making i
a u16.

Addresses-Coverity: ("Infinite loop")
Fixes: b36c598c999c ("ice: Updates to Tx scheduler code")
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

qede: fix NULL pointer deref in __qede_remove()

While rebooting the system with SR-IOV vfs enabled leads
to below crash due to recurrence of __qede_remove() on the VF
devices (first from .shutdown() flow of the VF itself and
another from PF's .shutdown() flow executing pci_disable_sriov())

This patch adds a safeguard in __qede_remove() flow to fix this,
so that driver doesn't attempt to remove "already removed" devices.

[  194.360134] BUG: unable to handle kernel NULL pointer dereference at 00000000000008dc
[  194.360227] IP: [<ffffffffc03553c4>] __qede_remove+0x24/0x130 [qede]
[  194.360304] PGD 0
[  194.360325] Oops: 0000 [#1] SMP
[  194.360360] Modules linked in: tcp_lp fuse tun bridge stp llc devlink bonding ip_set nfnetlink ib_isert iscsi_target_mod ib_srpt target_core_mod ib_srp scsi_transport_srp scsi_tgt ib_ipoib ib_umad rpcrdma sunrpc rdma_ucm ib_uverbs ib_iser rdma_cm iw_cm ib_cm libiscsi scsi_transport_iscsi dell_smbios iTCO_wdt iTCO_vendor_support dell_wmi_descriptor dcdbas vfat fat pcc_cpufreq skx_edac intel_powerclamp coretemp intel_rapl iosf_mbi kvm_intel kvm irqbypass crc32_pclmul ghash_clmulni_intel aesni_intel lrw gf128mul glue_helper ablk_helper cryptd qedr ib_core pcspkr ses enclosure joydev ipmi_ssif sg i2c_i801 lpc_ich mei_me mei wmi ipmi_si ipmi_devintf ipmi_msghandler tpm_crb acpi_pad acpi_power_meter xfs libcrc32c sd_mod crc_t10dif crct10dif_generic crct10dif_pclmul crct10dif_common crc32c_intel mgag200
[  194.361044]  qede i2c_algo_bit drm_kms_helper qed syscopyarea sysfillrect nvme sysimgblt fb_sys_fops ttm nvme_core mpt3sas crc8 ptp drm pps_core ahci raid_class scsi_transport_sas libahci libata drm_panel_orientation_quirks nfit libnvdimm dm_mirror dm_region_hash dm_log dm_mod [last unloaded: ip_tables]
[  194.361297] CPU: 51 PID: 7996 Comm: reboot Kdump: loaded Not tainted 3.10.0-1062.el7.x86_64 #1
[  194.361359] Hardware name: Dell Inc. PowerEdge MX840c/0740HW, BIOS 2.4.6 10/15/2019
[  194.361412] task: ffff9cea9b360000 ti: ffff9ceabebdc000 task.ti: ffff9ceabebdc000
[  194.361463] RIP: 0010:[<ffffffffc03553c4>]  [<ffffffffc03553c4>] __qede_remove+0x24/0x130 [qede]
[  194.361534] RSP: 0018:ffff9ceabebdfac0  EFLAGS: 00010282
[  194.361570] RAX: 0000000000000000 RBX: ffff9cd013846098 RCX: 0000000000000000
[  194.361621] RDX: 0000000000000000 RSI: 0000000000000000 RDI: ffff9cd013846098
[  194.361668] RBP: ffff9ceabebdfae8 R08: 0000000000000000 R09: 0000000000000000
[  194.361715] R10: 00000000bfe14201 R11: ffff9ceabfe141e0 R12: 0000000000000000
[  194.361762] R13: ffff9cd013846098 R14: 0000000000000000 R15: ffff9ceab5e48000
[  194.361810] FS:  00007f799c02d880(0000) GS:ffff9ceacb0c0000(0000) knlGS:0000000000000000
[  194.361865] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[  194.361903] CR2: 00000000000008dc CR3: 0000001bdac76000 CR4: 00000000007607e0
[  194.361953] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
[  194.362002] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
[  194.362051] PKRU: 55555554
[  194.362073] Call Trace:
[  194.362109]  [<ffffffffc0355500>] qede_remove+0x10/0x20 [qede]
[  194.362180]  [<ffffffffb97d0f3e>] pci_device_remove+0x3e/0xc0
[  194.362240]  [<ffffffffb98b3c52>] __device_release_driver+0x82/0xf0
[  194.362285]  [<ffffffffb98b3ce3>] device_release_driver+0x23/0x30
[  194.362343]  [<ffffffffb97c86d4>] pci_stop_bus_device+0x84/0xa0
[  194.362388]  [<ffffffffb97c87e2>] pci_stop_and_remove_bus_device+0x12/0x20
[  194.362450]  [<ffffffffb97f153f>] pci_iov_remove_virtfn+0xaf/0x160
[  194.362496]  [<ffffffffb97f1aec>] sriov_disable+0x3c/0xf0
[  194.362534]  [<ffffffffb97f1bc3>] pci_disable_sriov+0x23/0x30
[  194.362599]  [<ffffffffc02f83c3>] qed_sriov_disable+0x5e3/0x650 [qed]
[  194.362658]  [<ffffffffb9622df6>] ? kfree+0x106/0x140
[  194.362709]  [<ffffffffc02cc0c0>] ? qed_free_stream_mem+0x70/0x90 [qed]
[  194.362754]  [<ffffffffb9622df6>] ? kfree+0x106/0x140
[  194.362803]  [<ffffffffc02cd659>] qed_slowpath_stop+0x1a9/0x1d0 [qed]
[  194.362854]  [<ffffffffc035544e>] __qede_remove+0xae/0x130 [qede]
[  194.362904]  [<ffffffffc03554e0>] qede_shutdown+0x10/0x20 [qede]
[  194.362956]  [<ffffffffb97cf90a>] pci_device_shutdown+0x3a/0x60
[  194.363010]  [<ffffffffb98b180b>] device_shutdown+0xfb/0x1f0
[  194.363066]  [<ffffffffb94b66c6>] kernel_restart_prepare+0x36/0x40
[  194.363107]  [<ffffffffb94b66e2>] kernel_restart+0x12/0x60
[  194.363146]  [<ffffffffb94b6959>] SYSC_reboot+0x229/0x260
[  194.363196]  [<ffffffffb95f200d>] ? handle_mm_fault+0x39d/0x9b0
[  194.363253]  [<ffffffffb942b621>] ? __switch_to+0x151/0x580
[  194.363304]  [<ffffffffb9b7ec28>] ? __schedule+0x448/0x9c0
[  194.363343]  [<ffffffffb94b69fe>] SyS_reboot+0xe/0x10
[  194.363387]  [<ffffffffb9b8bede>] system_call_fastpath+0x25/0x2a
[  194.363430] Code: f9 e9 37 ff ff ff 90 0f 1f 44 00 00 55 48 89 e5 41 57 41 56 41 55 4c 8d af 98 00 00 00 41 54 4c 89 ef 41 89 f4 53 e8 4c e4 55 f9 <80> b8 dc 08 00 00 01 48 89 c3 4c 8d b8 c0 08 00 00 4c 8b b0 c0
[  194.363712] RIP  [<ffffffffc03553c4>] __qede_remove+0x24/0x130 [qede]
[  194.363764]  RSP <ffff9ceabebdfac0>
[  194.363791] CR2: 00000000000008dc

Signed-off-by: Manish Chopra <manishc@marvell.com>
Signed-off-by: Ariel Elior <aelior@marvell.com>
Signed-off-by: Sudarsana Kalluru <skalluru@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: fix data-race in neigh_event_send()

KCSAN reported the following data-race [1]

The fix will also prevent the compiler from optimizing out
the condition.

[1]

BUG: KCSAN: data-race in neigh_resolve_output / neigh_resolve_output

write to 0xffff8880a41dba78 of 8 bytes by interrupt on cpu 1:
 neigh_event_send include/net/neighbour.h:443 [inline]
 neigh_resolve_output+0x78/0x480 net/core/neighbour.c:1474
 neigh_output include/net/neighbour.h:511 [inline]
 ip_finish_output2+0x4af/0xe40 net/ipv4/ip_output.c:228
 __ip_finish_output net/ipv4/ip_output.c:308 [inline]
 __ip_finish_output+0x23a/0x490 net/ipv4/ip_output.c:290
 ip_finish_output+0x41/0x160 net/ipv4/ip_output.c:318
 NF_HOOK_COND include/linux/netfilter.h:294 [inline]
 ip_output+0xdf/0x210 net/ipv4/ip_output.c:432
 dst_output include/net/dst.h:436 [inline]
 ip_local_out+0x74/0x90 net/ipv4/ip_output.c:125
 __ip_queue_xmit+0x3a8/0xa40 net/ipv4/ip_output.c:532
 ip_queue_xmit+0x45/0x60 include/net/ip.h:237
 __tcp_transmit_skb+0xe81/0x1d60 net/ipv4/tcp_output.c:1169
 tcp_transmit_skb net/ipv4/tcp_output.c:1185 [inline]
 __tcp_retransmit_skb+0x4bd/0x15f0 net/ipv4/tcp_output.c:2976
 tcp_retransmit_skb+0x36/0x1a0 net/ipv4/tcp_output.c:2999
 tcp_retransmit_timer+0x719/0x16d0 net/ipv4/tcp_timer.c:515
 tcp_write_timer_handler+0x42d/0x510 net/ipv4/tcp_timer.c:598
 tcp_write_timer+0xd1/0xf0 net/ipv4/tcp_timer.c:618

read to 0xffff8880a41dba78 of 8 bytes by interrupt on cpu 0:
 neigh_event_send include/net/neighbour.h:442 [inline]
 neigh_resolve_output+0x57/0x480 net/core/neighbour.c:1474
 neigh_output include/net/neighbour.h:511 [inline]
 ip_finish_output2+0x4af/0xe40 net/ipv4/ip_output.c:228
 __ip_finish_output net/ipv4/ip_output.c:308 [inline]
 __ip_finish_output+0x23a/0x490 net/ipv4/ip_output.c:290
 ip_finish_output+0x41/0x160 net/ipv4/ip_output.c:318
 NF_HOOK_COND include/linux/netfilter.h:294 [inline]
 ip_output+0xdf/0x210 net/ipv4/ip_output.c:432
 dst_output include/net/dst.h:436 [inline]
 ip_local_out+0x74/0x90 net/ipv4/ip_output.c:125
 __ip_queue_xmit+0x3a8/0xa40 net/ipv4/ip_output.c:532
 ip_queue_xmit+0x45/0x60 include/net/ip.h:237
 __tcp_transmit_skb+0xe81/0x1d60 net/ipv4/tcp_output.c:1169
 tcp_transmit_skb net/ipv4/tcp_output.c:1185 [inline]
 __tcp_retransmit_skb+0x4bd/0x15f0 net/ipv4/tcp_output.c:2976
 tcp_retransmit_skb+0x36/0x1a0 net/ipv4/tcp_output.c:2999
 tcp_retransmit_timer+0x719/0x16d0 net/ipv4/tcp_timer.c:515
 tcp_write_timer_handler+0x42d/0x510 net/ipv4/tcp_timer.c:598

Reported by Kernel Concurrency Sanitizer on:
CPU: 0 PID: 0 Comm: swapper/0 Not tainted 5.4.0-rc3+ #0
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011

Signed-off-by: Eric Dumazet <edumazet@google.com>
Reported-by: syzbot <syzkaller@googlegroups.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Merge tag 'pwm/for-5.4-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/thierry.reding/linux-pwm

Pull pwm fix from Thierry Reding:
 "One more fix to keep a reference to the driver's module as long as
  there are users of the PWM exposed by the driver"

* tag 'pwm/for-5.4-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/thierry.reding/linux-pwm:
  pwm: bcm-iproc: Prevent unloading the driver module while in use

sched: Fix pick_next_task() vs 'change' pattern race

Commit 67692435c411 ("sched: Rework pick_next_task() slow-path")
inadvertly introduced a race because it changed a previously
unexplored dependency between dropping the rq->lock and
sched_class::put_prev_task().

The comments about dropping rq->lock, in for example
newidle_balance(), only mentions the task being current and ->on_cpu
being set. But when we look at the 'change' pattern (in for example
sched_setnuma()):

queued = task_on_rq_queued(p); /* p->on_rq == TASK_ON_RQ_QUEUED */
running = task_current(rq, p); /* rq->curr == p */

if (queued)
dequeue_task(...);
if (running)
put_prev_task(...);

/* change task properties */

if (queued)
enqueue_task(...);
if (running)
set_next_task(...);

It becomes obvious that if we do this after put_prev_task() has
already been called on @p, things go sideways. This is exactly what
the commit in question allows to happen when it does:

prev->sched_class->put_prev_task(rq, prev, rf);
if (!rq->nr_running)
newidle_balance(rq, rf);

The newidle_balance() call will drop rq->lock after we've called
put_prev_task() and that allows the above 'change' pattern to
interleave and mess up the state.

Furthermore, it turns out we lost the RT-pull when we put the last DL
task.

Fix both problems by extracting the balancing from put_prev_task() and
doing a multi-class balance() pass before put_prev_task().

Fixes: 67692435c411 ("sched: Rework pick_next_task() slow-path")
Reported-by: Quentin Perret <qperret@google.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Tested-by: Quentin Perret <qperret@google.com>
Tested-by: Valentin Schneider <valentin.schneider@arm.com>

sched/core: Fix compilation error when cgroup not selected

When cgroup is disabled the following compilation error was hit

kernel/sched/core.c: In function ‘uclamp_update_active_tasks’:
kernel/sched/core.c:1081:23: error: storage size of ‘it’ isn’t known
  struct css_task_iter it;
       ^~
kernel/sched/core.c:1084:2: error: implicit declaration of function ‘css_task_iter_start’; did you mean ‘__sg_page_iter_start’? [-Werror=implicit-function-declaration]
  css_task_iter_start(css, 0, &it);
  ^~~~~~~~~~~~~~~~~~~
  __sg_page_iter_start
kernel/sched/core.c:1085:14: error: implicit declaration of function ‘css_task_iter_next’; did you mean ‘__sg_page_iter_next’? [-Werror=implicit-function-declaration]
  while ((p = css_task_iter_next(&it))) {
      ^~~~~~~~~~~~~~~~~~
      __sg_page_iter_next
kernel/sched/core.c:1091:2: error: implicit declaration of function ‘css_task_iter_end’; did you mean ‘get_task_cred’? [-Werror=implicit-function-declaration]
  css_task_iter_end(&it);
  ^~~~~~~~~~~~~~~~~
  get_task_cred
kernel/sched/core.c:1081:23: warning: unused variable ‘it’ [-Wunused-variable]
  struct css_task_iter it;
       ^~
cc1: some warnings being treated as errors
make[2]: *** [kernel/sched/core.o] Error 1

Fix by protetion uclamp_update_active_tasks() with
CONFIG_UCLAMP_TASK_GROUP

Fixes: babbe170e053 ("sched/uclamp: Update CPU's refcount on TG's clamp changes")
Reported-by: Randy Dunlap <rdunlap@infradead.org>
Signed-off-by: Qais Yousef <qais.yousef@arm.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Tested-by: Randy Dunlap <rdunlap@infradead.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Vincent Guittot <vincent.guittot@linaro.org>
Cc: Patrick Bellasi <patrick.bellasi@matbug.net>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Dietmar Eggemann <dietmar.eggemann@arm.com>
Cc: Juri Lelli <juri.lelli@redhat.com>
Cc: Ben Segall <bsegall@google.com>
Link: https://lkml.kernel.org/r/20191105112212.596-1-qais.yousef@arm.com

cgroup,writeback: don't switch wbs immediately on dead wbs if the memcg is dead

cgroup writeback tries to refresh the associated wb immediately if the
current wb is dead.  This is to avoid keeping issuing IOs on the stale
wb after memcg - blkcg association has changed (ie. when blkcg got
disabled / enabled higher up in the hierarchy).

Unfortunately, the logic gets triggered spuriously on inodes which are
associated with dead cgroups.  When the logic is triggered on dead
cgroups, the attempt fails only after doing quite a bit of work
allocating and initializing a new wb.

While c3aab9a0bd91 ("mm/filemap.c: don't initiate writeback if mapping
has no dirty pages") alleviated the issue significantly as it now only
triggers when the inode has dirty pages.  However, the condition can
still be triggered before the inode is switched to a different cgroup
and the logic simply doesn't make sense.

Skip the immediate switching if the associated memcg is dying.

This is a simplified version of the following two patches:

 * https://lore.kernel.org/linux-mm/20190513183053.GA73423@dennisz-mbp/
 * http://lkml.kernel.org/r/156355839560.2063.5265687291430814589.stgit@buzz

Cc: Konstantin Khlebnikov <khlebnikov@yandex-team.ru>
Fixes: e8a7abf5a5bd ("writeback: disassociate inodes from dying bdi_writebacks")
Acked-by: Dennis Zhou <dennis@kernel.org>
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jens Axboe <axboe@kernel.dk>

Merge tag 'ceph-for-5.4-rc7' of git://github.com/ceph/ceph-client

Pull ceph fixes from Ilya Dryomov:
 "Some late-breaking dentry handling fixes from Al and Jeff, a patch to
  further restrict copy_file_range() to avoid potential data corruption
  from Luis and a fix for !CONFIG_CEPH_FSCACHE kernels.

  Everything but the fscache fix is marked for stable"

* tag 'ceph-for-5.4-rc7' of git://github.com/ceph/ceph-client:
  ceph: return -EINVAL if given fsc mount option on kernel w/o support
  ceph: don't allow copy_file_range when stripe_count != 1
  ceph: don't try to handle hashed dentries in non-O_CREAT atomic_open
  ceph: add missing check in d_revalidate snapdir handling
  ceph: fix RCU case handling in ceph_d_revalidate()
  ceph: fix use-after-free in __ceph_remove_cap()

vsock/virtio: fix sock refcnt holding during the shutdown

The "42f5cda5eaf4" commit rightly set SOCK_DONE on peer shutdown,
but there is an issue if we receive the SHUTDOWN(RDWR) while the
virtio_transport_close_timeout() is scheduled.
In this case, when the timeout fires, the SOCK_DONE is already
set and the virtio_transport_close_timeout() will not call
virtio_transport_reset() and virtio_transport_do_close().
This causes that both sockets remain open and will never be released,
preventing the unloading of [virtio|vhost]_transport modules.

This patch fixes this issue, calling virtio_transport_reset() and
virtio_transport_do_close() when we receive the SHUTDOWN(RDWR)
and there is nothing left to read.

Fixes: 42f5cda5eaf4 ("vsock/virtio: set SOCK_DONE on peer shutdown")
Cc: Stephen Barber <smbarber@chromium.org>
Signed-off-by: Stefano Garzarella <sgarzare@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Merge tag 'mac80211-for-net-2019-11-08' of git://git.kernel.org/pub/scm/linux/kernel/git/jberg/mac80211

Johannes Berg says:

====================
Three small fixes:
 * we hit a failure path bug related to
   ieee80211_txq_setup_flows()
 * also use kvmalloc() to make that less likely
 * fix a timing value shortly after boot (during
   INITIAL_JIFFIES)
====================

Signed-off-by: David S. Miller <davem@davemloft.net>

net: ethernet: octeon_mgmt: Account for second possible VLAN header

Octeon's input ring-buffer entry has 14 bits-wide size field, so to account
for second possible VLAN header max_mtu must be further reduced.

Fixes: 109cc16526c6d ("ethernet/cavium: use core min/max MTU checking")
Signed-off-by: Alexander Sverdlin <alexander.sverdlin@nokia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Merge tag 'modules-for-v5.4-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/jeyu/linux

Pull modules fix from Jessica Yu:
 "Fix `make nsdeps` for modules composed of multiple source files.

  Since $mod_source_files was not in quotes in the call to
  generate_deps_for_ns(), not all the source files for a module were
  being passed to spatch"

* tag 'modules-for-v5.4-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/jeyu/linux:
  scripts/nsdeps: make sure to pass all module source files to spatch

Merge tag 'arm64-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux

Pull arm64 fix from Will Deacon:
 "Fix pte_same() to avoid getting stuck on write fault.

  This single arm64 fix is a revert of 747a70e60b72 ("arm64: Fix
  copy-on-write referencing in HugeTLB"), not because that patch was
  wrong, but because it was broken by aa57157be69f ("arm64: Ensure
  VM_WRITE|VM_SHARED ptes are clean by default") which we merged in
  -rc6.

  We spotted the issue in Android (AOSP), where one of the JIT threads
  gets stuck on a write fault during boot because the faulting pte is
  marked as PTE_DIRTY | PTE_WRITE | PTE_RDONLY and the fault handler
  decides that there's nothing to do thanks to pte_same() masking out
  PTE_RDONLY.

  Thanks to John Stultz for reporting this and testing this so quickly,
  and to Steve Capper for confirming that the HugeTLB tests continue to
  pass"

* tag 'arm64-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux:
  arm64: Do not mask out PTE_RDONLY in pte_same()

pwm: bcm-iproc: Prevent unloading the driver module while in use

The owner member of struct pwm_ops must be set to THIS_MODULE to
increase the reference count of the module such that the module cannot
be removed while its code is in use.

Fixes: daa5abc41c80 ("pwm: Add support for Broadcom iProc PWM controller")
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Thierry Reding <thierry.reding@gmail.com>

Merge tag 'xarray-5.4' of git://git.infradead.org/users/willy/linux-dax

Pull XArray fixes from Matthew Wilcox:
 "These all fix various bugs, some of which people have tripped over and
  some of which have been caught by automatic tools"

* tag 'xarray-5.4' of git://git.infradead.org/users/willy/linux-dax:
  idr: Fix idr_alloc_u32 on 32-bit systems
  idr: Fix integer overflow in idr_for_each_entry
  radix tree: Remove radix_tree_iter_find
  idr: Fix idr_get_next_ul race with idr_remove
  XArray: Fix xas_next() with a single entry at 0

Merge tag 'pm-5.4-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm

Pull power management fix from Rafael Wysocki:
 "Fix an 'unchecked MSR access' warning in the intel_pstate cpufreq
  driver (Srinivas Pandruvada)"

* tag 'pm-5.4-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm:
  cpufreq: intel_pstate: Fix invalid EPB setting

Merge tag 'sound-5.4-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound

Pull sound fixes from Takashi Iwai:
 "It became a bit largish, but all small and good for 5.4:

   - A regression fix of ALSA timer code bug that sneaked in by a recent
     cleanup; never trust innocent-looking guys...

   - Fix for compress API max size check signedness

   - Fixes in HD-audio: CA0132 work stall, Intel Tigerlake HDMI

   - A few fixes for SOF: memory leak, sanity-check and build fixes

   - A collection of device-specific fixes: firewire, rockchip, ASoC
     HDMI, rsnd, ASoC HDA, stm32, TI, kirkwood, msm, max98373"

* tag 'sound-5.4-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound:
  ALSA: timer: Fix incorrectly assigned timer instance
  ASoC: SOF: topology: Fix bytes control size checks
  ALSA: hda: hdmi - add Tigerlake support
  ASoC: max98373: replace gpio_request with devm_gpio_request
  ASoC: stm32: sai: add restriction on mmap support
  ALSA: hda/ca0132 - Fix possible workqueue stall
  ASoC: hdac_hda: fix race in device removal
  ALSA: bebob: fix to detect configured source of sampling clock for Focusrite Saffire Pro i/o series
  ASoC: rockchip: rockchip_max98090: Enable SHDN to fix headset detection
  ASoC: ti: sdma-pcm: Add back the flags parameter for non standard dma names
  ASoC: SOF: ipc: Fix memory leak in sof_set_get_large_ctrl_data
  ASoC: SOF: Fix memory leak in sof_dfsentry_write
  ASoC: SOF: Intel: hda-stream: fix the CONFIG_ prefix missing
  ASoC: kirkwood: fix device remove ordering
  ASoC: rsnd: dma: fix SSI9 4/5/6/7 busif dma address
  ASoC: hdmi-codec: drop mutex locking again
  ASoC: kirkwood: fix external clock probe defer
  ASoC: compress: fix unsigned integer overflow check
  ASoC: msm8916-wcd-analog: Fix RX1 selection in RDAC2 MUX

Merge tag 'drm-fixes-2019-11-08' of git://anongit.freedesktop.org/drm/drm

Pull drm fixes from Dave Airlie:
 "Weekly fixes for drm: amdgpu has a few but they are pretty scattered
  fixes, the fbdev one is a build regression fix that we didn't want to
  risk leaving out, otherwise a couple of i915, one radeon and a core
  atomic fix.

  core:
   - add missing documentation for GEM shmem madvise helpers
   - Fix for a state dereference in atomic self-refresh helpers

  fbdev:
   - One compilation fix for c2p fbdev helpers

  amdgpu:
   - Fix navi14 display issue root cause and revert workaround
   - GPU reset scheduler interaction fix
   - Fix fan boost on multi-GPU
   - Gfx10 and sdma5 fixes for navi
   - GFXOFF fix for renoir
   - Add navi14 PCI ID
   - GPUVM fix for arcturus

  radeon:
   - Port an SI power fix from amdgpu

  i915:
   - Fix HPD poll to avoid kworker consuming a lot of cpu cycles.
   - Do not use TBT type for non Type-C ports"

* tag 'drm-fixes-2019-11-08' of git://anongit.freedesktop.org/drm/drm:
  drm/radeon: fix si_enable_smc_cac() failed issue
  drm/amdgpu/renoir: move gfxoff handling into gfx9 module
  drm/amdgpu: add warning for GRBM 1-cycle delay issue in gfx9
  drm/amdgpu: add dummy read by engines for some GCVM status registers in gfx10
  drm/amdgpu: register gpu instance before fan boost feature enablment
  drm/amd/swSMU: fix smu workload bit map error
  drm/shmem: Add docbook comments for drm_gem_shmem_object madvise fields
  drm/amdgpu: add navi14 PCI ID
  Revert "drm/amd/display: setting the DIG_MODE to the correct value."
  drm/amd/display: Add ENGINE_ID_DIGD condition check for Navi14
  drm/amdgpu: dont schedule jobs while in reset
  drm/amdgpu/arcturus: properly set BANK_SELECT and FRAGMENT_SIZE
  drm/atomic: fix self-refresh helpers crtc state dereference
  drm/i915/dp: Do not switch aux to TBT mode for non-TC ports
  drm/i915: Avoid HPD poll detect triggering a new detect cycle
  fbdev: c2p: Fix link failure on non-inlining

Merge tag 'clk-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/clk/linux

Pull clk fixes from Stephen Boyd:
 "Fixes for various clk driver issues that happened because of code we
  merged this merge window.

  The Amlogic driver was missing some flags causing rates to be rounded
  improperly or clk_set_rate() to fail. The Samsung driver wasn't
  freeing everything on error paths and improperly saving/restoring PLL
  state across suspend/resume. The at91 driver was calling msleep() too
  early when scheduling hadn't started, so we put in place a quick
  solution until we can handle this sort of problem in the core
  framework.

  There were also problems with the Allwinner driver and operator
  precedence being incorrect causing subtle bugs. Finally, the TI driver
  was duplicating aliases and not delaying long enough leading to some
  unexpected timeouts"

* tag 'clk-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/clk/linux:
  clk: ti: clkctrl: Fix failed to enable error with double udelay timeout
  clk: ti: dra7-atl-clock: Remove ti_clk_add_alias call
  clk: sunxi-ng: a80: fix the zero'ing of bits 16 and 18
  clk: sunxi: Fix operator precedence in sunxi_divs_clk_setup
  clk: ast2600: Fix enabling of clocks
  clk: at91: avoid sleeping early
  clk: imx8m: Use SYS_PLL1_800M as intermediate parent of CLK_ARM
  clk: samsung: exynos5420: Preserve PLL configuration during suspend/resume
  clk: samsung: exynos542x: Move G3D subsystem clocks to its sub-CMU
  clk: samsung: exynos5433: Fix error paths
  clk: at91: sam9x60: fix programmable clock
  clk: meson: g12a: set CLK_MUX_ROUND_CLOSEST on the cpu clock muxes
  clk: meson: g12a: fix cpu clock rate setting
  clk: meson: gxbb: let sar_adc_clk_div set the parent clock rate

block: drbd: remove a stray unlock in __drbd_send_protocol()

There are two callers of this function and they both unlock the mutex so
this ends up being a double unlock.

Fixes: 44ed167da748 ("drbd: rcu_read_lock() and rcu_dereference() for tconn->net_conf")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>

cpufreq: intel_pstate: Fix invalid EPB setting

The max value of EPB can only be 0x0F. Attempting to set more than that
triggers an "unchecked MSR access error" warning which happens in
intel_pstate_hwp_force_min_perf() called via cpufreq stop_cpu().

However, it is not even necessary to touch the EPB from intel_pstate,
because it is restored on every CPU online by the intel_epb.c code,
so let that code do the right thing and drop the redundant (and
incorrect) EPB update from intel_pstate.

Fixes: af3b7379e2d70 ("cpufreq: intel_pstate: Force HWP min perf before offline")
Reported-by: Qian Cai <cai@lca.pw>
Cc: 5.2+ <stable@vger.kernel.org> # 5.2+
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
[ rjw: Changelog ]
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>

mac80211: fix station inactive_time shortly after boot

In the first 5 minutes after boot (time of INITIAL_JIFFIES),
ieee80211_sta_last_active() returns zero if last_ack is zero. This
leads to "inactive time" showing jiffies_to_msecs(jiffies).

 # iw wlan0 station get fc:ec:da:64:a6:dd
 Station fc:ec:da:64:a6:dd (on wlan0)
inactive time: 4294894049 ms
.
.
connected time: 70 seconds

Fix by returning last_rx if last_ack == 0.

Signed-off-by: Ahmed Zaki <anzaki@gmail.com>
Link: https://lore.kernel.org/r/20191031121243.27694-1-anzaki@gmail.com
Signed-off-by: Johannes Berg <johannes.berg@intel.com>

net/fq_impl: Switch to kvmalloc() for memory allocation

The FQ implementation used by mac80211 allocates memory using kmalloc(),
which can fail; and Johannes reported that this actually happens in
practice.

To avoid this, switch the allocation to kvmalloc() instead; this also
brings fq_impl in line with all the FQ qdiscs.

Fixes: 557fc4a09803 ("fq: add fair queuing framework")
Reported-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com>
Link: https://lore.kernel.org/r/20191105155750.547379-1-toke@redhat.com
Signed-off-by: Johannes Berg <johannes.berg@intel.com>

mac80211: fix ieee80211_txq_setup_flows() failure path

If ieee80211_txq_setup_flows() fails, we don't clean up LED
state properly, leading to crashes later on, fix that.

Fixes: dc8b274f0952 ("mac80211: Move up init of TXQs")
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Acked-by: Toke Høiland-Jørgensen <toke@toke.dk>
Link: https://lore.kernel.org/r/20191105154110.1ccf7112ba5d.I0ba865792446d051867b33153be65ce6b063d98c@changeid
Signed-off-by: Johannes Berg <johannes.berg@intel.com>

Merge tag 'drm-fixes-5.4-2019-11-06' of git://people.freedesktop.org/~agd5f/linux into drm-fixes

drm-fixes-5.4-2019-11-06:

amdgpu:
- Fix navi14 display issue root cause and revert workaround
- GPU reset scheduler interaction fix
- Fix fan boost on multi-GPU
- Gfx10 and sdma5 fixes for navi
- GFXOFF fix for renoir
- Add navi14 PCI ID
- GPUVM fix for arcturus

radeon:
- Port an SI power fix from amdgpu

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Alex Deucher <alexdeucher@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20191107032241.1021217-1-alexander.deucher@amd.com

Merge tag 'drm-intel-fixes-2019-11-06' of git://anongit.freedesktop.org/drm/drm-intel into drm-fixes

- Fix HPD poll to avoid kworker consuming a lot of cpu cycles.
- Do not use TBT type for non Type-C ports.

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Rodrigo Vivi <rodrigo.vivi@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20191106213958.GA16525@intel.com

Merge tag 'drm-misc-fixes-2019-11-07-1' of git://anongit.freedesktop.org/drm/drm-misc into drm-fixes

 - Some new documentation for GEM shmem madvise helpers
 - Fix for a state dereference in atomic self-refresh helpers
 - One compilation fix for c2p fbdev helpers

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Maxime Ripard <mripard@kernel.org>
Link: https://patchwork.freedesktop.org/patch/msgid/20191107082215.GA34850@gilmour.lan

ipv4: Fix table id reference in fib_sync_down_addr

Hendrik reported routes in the main table using source address are not
removed when the address is removed. The problem is that fib_sync_down_addr
does not account for devices in the default VRF which are associated
with the main table. Fix by updating the table id reference.

Fixes: 5a56a0b3a45d ("net: Don't delete routes in different VRFs")
Reported-by: Hendrik Donner <hd@os-cillation.de>
Signed-off-by: David Ahern <dsahern@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

ipv6: fixes rt6_probe() and fib6_nh->last_probe init

While looking at a syzbot KCSAN report [1], I found multiple
issues in this code :

1) fib6_nh->last_probe has an initial value of 0.

   While probably okay on 64bit kernels, this causes an issue
   on 32bit kernels since the time_after(jiffies, 0 + interval)
   might be false ~24 days after boot (for HZ=1000)

2) The data-race found by KCSAN
   I could use READ_ONCE() and WRITE_ONCE(), but we also can
   take the opportunity of not piling-up too many rt6_probe_deferred()
   works by using instead cmpxchg() so that only one cpu wins the race.

[1]
BUG: KCSAN: data-race in find_match / find_match

write to 0xffff8880bb7aabe8 of 8 bytes by interrupt on cpu 1:
 rt6_probe net/ipv6/route.c:663 [inline]
 find_match net/ipv6/route.c:757 [inline]
 find_match+0x5bd/0x790 net/ipv6/route.c:733
 __find_rr_leaf+0xe3/0x780 net/ipv6/route.c:831
 find_rr_leaf net/ipv6/route.c:852 [inline]
 rt6_select net/ipv6/route.c:896 [inline]
 fib6_table_lookup+0x383/0x650 net/ipv6/route.c:2164
 ip6_pol_route+0xee/0x5c0 net/ipv6/route.c:2200
 ip6_pol_route_output+0x48/0x60 net/ipv6/route.c:2452
 fib6_rule_lookup+0x3d6/0x470 net/ipv6/fib6_rules.c:117
 ip6_route_output_flags_noref+0x16b/0x230 net/ipv6/route.c:2484
 ip6_route_output_flags+0x50/0x1a0 net/ipv6/route.c:2497
 ip6_dst_lookup_tail+0x25d/0xc30 net/ipv6/ip6_output.c:1049
 ip6_dst_lookup_flow+0x68/0x120 net/ipv6/ip6_output.c:1150
 inet6_csk_route_socket+0x2f7/0x420 net/ipv6/inet6_connection_sock.c:106
 inet6_csk_xmit+0x91/0x1f0 net/ipv6/inet6_connection_sock.c:121
 __tcp_transmit_skb+0xe81/0x1d60 net/ipv4/tcp_output.c:1169
 tcp_transmit_skb net/ipv4/tcp_output.c:1185 [inline]
 tcp_xmit_probe_skb+0x19b/0x1d0 net/ipv4/tcp_output.c:3735

read to 0xffff8880bb7aabe8 of 8 bytes by interrupt on cpu 0:
 rt6_probe net/ipv6/route.c:657 [inline]
 find_match net/ipv6/route.c:757 [inline]
 find_match+0x521/0x790 net/ipv6/route.c:733
 __find_rr_leaf+0xe3/0x780 net/ipv6/route.c:831
 find_rr_leaf net/ipv6/route.c:852 [inline]
 rt6_select net/ipv6/route.c:896 [inline]
 fib6_table_lookup+0x383/0x650 net/ipv6/route.c:2164
 ip6_pol_route+0xee/0x5c0 net/ipv6/route.c:2200
 ip6_pol_route_output+0x48/0x60 net/ipv6/route.c:2452
 fib6_rule_lookup+0x3d6/0x470 net/ipv6/fib6_rules.c:117
 ip6_route_output_flags_noref+0x16b/0x230 net/ipv6/route.c:2484
 ip6_route_output_flags+0x50/0x1a0 net/ipv6/route.c:2497
 ip6_dst_lookup_tail+0x25d/0xc30 net/ipv6/ip6_output.c:1049
 ip6_dst_lookup_flow+0x68/0x120 net/ipv6/ip6_output.c:1150
 inet6_csk_route_socket+0x2f7/0x420 net/ipv6/inet6_connection_sock.c:106
 inet6_csk_xmit+0x91/0x1f0 net/ipv6/inet6_connection_sock.c:121
 __tcp_transmit_skb+0xe81/0x1d60 net/ipv4/tcp_output.c:1169

Reported by Kernel Concurrency Sanitizer on:
CPU: 0 PID: 18894 Comm: udevd Not tainted 5.4.0-rc3+ #0
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011

Fixes: cc3a86c802f0 ("ipv6: Change rt6_probe to take a fib6_nh")
Fixes: f547fac624be ("ipv6: rate-limit probes for neighbourless routes")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reported-by: syzbot <syzkaller@googlegroups.com>
Reviewed-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: hns: Fix the stray netpoll locks causing deadlock in NAPI path

This patch fixes the problem of the spin locks, originally
meant for the netpoll path of hns driver, causing deadlock in
the normal NAPI poll path. The issue happened due to the presence
of the stray leftover spin lock code related to the netpoll,
whose support was earlier removed from the HNS[1], got activated
due to enabling of NET_POLL_CONTROLLER switch.

Earlier background:
The netpoll handling code originally had this bug(as identified
by Marc Zyngier[2]) of wrong spin lock API being used which did
not disable the interrupts and hence could cause locking issues.
i.e. if the lock were first acquired in context to thread like
'ip' util and this lock if ever got later acquired again in
context to the interrupt context like TX/RX (Interrupts could
always pre-empt the lock holding task and acquire the lock again)
and hence could cause deadlock.

Proposed Solution:
1. If the netpoll was enabled in the HNS driver, which is not
   right now, we could have simply used spin_[un]lock_irqsave()
2. But as netpoll is disabled, therefore, it is best to get rid
   of the existing locks and stray code for now. This should
   solve the problem reported by Marc.

[1] https://git.kernel.org/torvalds/c/4bd2c03be7
[2] https://patchwork.ozlabs.org/patch/1189139/

Fixes: 4bd2c03be707 ("net: hns: remove ndo_poll_controller")
Cc: lipeng <lipeng321@huawei.com>
Cc: Yisen Zhuang <yisen.zhuang@huawei.com>
Cc: Eric Dumazet <edumazet@google.com>
Cc: David S. Miller <davem@davemloft.net>
Reported-by: Marc Zyngier <maz@kernel.org>
Acked-by: Marc Zyngier <maz@kernel.org>
Tested-by: Marc Zyngier <maz@kernel.org>
Signed-off-by: Salil Mehta <salil.mehta@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: usb: qmi_wwan: add support for DW5821e with eSIM support

Exactly same layout as the default DW5821e module, just a different
vid/pid.

The QMI interface is exposed in USB configuration #1:

P:  Vendor=413c ProdID=81e0 Rev=03.18
S:  Manufacturer=Dell Inc.
S:  Product=DW5821e-eSIM Snapdragon X20 LTE
S:  SerialNumber=0123456789ABCDEF
C:  #Ifs= 6 Cfg#= 1 Atr=a0 MxPwr=500mA
I:  If#=0x0 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=ff Prot=ff Driver=qmi_wwan
I:  If#=0x1 Alt= 0 #EPs= 1 Cls=03(HID  ) Sub=00 Prot=00 Driver=usbhid
I:  If#=0x2 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=00 Prot=00 Driver=option
I:  If#=0x3 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=00 Prot=00 Driver=option
I:  If#=0x4 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=00 Prot=00 Driver=option
I:  If#=0x5 Alt= 0 #EPs= 2 Cls=ff(vend.) Sub=ff Prot=ff Driver=option

Signed-off-by: Aleksander Morgado <aleksander@aleksander.es>
Signed-off-by: David S. Miller <davem@davemloft.net>

CDC-NCM: handle incomplete transfer of MTU

A malicious device may give half an answer when asked
for its MTU. The driver will proceed after this with
a garbage MTU. Anything but a complete answer must be treated
as an error.

V2: used sizeof as request by Alexander

Reported-and-tested-by: syzbot+0631d878823ce2411636@syzkaller.appspotmail.com
Signed-off-by: Oliver Neukum <oneukum@suse.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

nfc: netlink: fix double device reference drop

The function nfc_put_device(dev) is called twice to drop the reference
to dev when there is no associated local llcp. Remove one of them to fix
the bug.

Fixes: 52feb444a903 ("NFC: Extend netlink interface for LTO, RW, and MIUX parameters support")
Fixes: d9b8d8e19b07 ("NFC: llcp: Service Name Lookup netlink interface")
Signed-off-by: Pan Bian <bianpan2016@163.com>
Reviewed-by: Johan Hovold <johan@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/hid/hid

Pull HID fixes from Jiri Kosina:
 "Two fixes for the HID subsystem:

   - regression fix for i2c-hid power management (Hans de Goede)

   - signed vs unsigned API fix for Wacom driver (Jason Gerecke)"

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/hid/hid:
  HID: wacom: generic: Treat serial number and related fields as unsigned
  HID: i2c-hid: Send power-on command after reset

ceph: return -EINVAL if given fsc mount option on kernel w/o support

If someone requests fscache on the mount, and the kernel doesn't
support it, it should fail the mount.

[ Drop ceph prefix -- it's provided by pr_err. ]

Signed-off-by: Jeff Layton <jlayton@kernel.org>
Reviewed-by: Ilya Dryomov <idryomov@gmail.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>

staging: Fix error return code in vboxsf_fill_super()

Fix to return negative error code -ENOMEM from the error handling
case instead of 0, as done elsewhere in this function.

Fixes: df4028658f9d ("staging: Add VirtualBox guest shared folder (vboxsf) support")
Signed-off-by: Wei Yongjun <weiyongjun1@huawei.com>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Link: https://lore.kernel.org/r/20191106115954.114678-1-weiyongjun1@huawei.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: vboxsf: fix dereference of pointer dentry before it is null checked

Currently the pointer dentry is being dereferenced before it is
being null checked.  Fix this by only dereferencing dentry once
we know it is not null.

Addresses-Coverity: ("Dereference before null check")
Fixes: df4028658f9d ("staging: Add VirtualBox guest shared folder (vboxsf) support")
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Link: https://lore.kernel.org/r/20191105175108.79824-1-colin.king@canonical.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: vboxsf: Remove unused including <linux/version.h>

Remove including <linux/version.h> that don't need it.

Signed-off-by: YueHaibing <yuehaibing@huawei.com>
Link: https://lore.kernel.org/r/20191107015923.100013-1-yuehaibing@huawei.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

perf report: Sort by sampled cycles percent per block for tui

Previous patch has implemented a new option "--total-cycles".  But only
stdio mode is supported.

This patch supports the tui mode and support '--percent-limit'.

For example,

 perf record -b ./div
 perf report --total-cycles --percent-limit 1

 # Samples: 2753248 of event 'cycles'
 Sampled Cycles%  Sampled Cycles  Avg Cycles%  Avg Cycles                                              [Program Block Range]         Shared Object
          26.04%            2.8M        0.40%          18                                             [div.c:42 -> div.c:39]                   div
          15.17%            1.2M        0.16%           7                                 [random_r.c:357 -> random_r.c:380]          libc-2.27.so
           5.11%          402.0K        0.04%           2                                             [div.c:27 -> div.c:28]                   div
           4.87%          381.6K        0.04%           2                                     [random.c:288 -> random.c:291]          libc-2.27.so
           4.53%          381.0K        0.04%           2                                             [div.c:40 -> div.c:40]                   div
           3.85%          300.9K        0.02%           1                                             [div.c:22 -> div.c:25]                   div
           3.08%          241.1K        0.02%           1                                           [rand.c:26 -> rand.c:27]          libc-2.27.so
           3.06%          240.0K        0.02%           1                                     [random.c:291 -> random.c:291]          libc-2.27.so
           2.78%          215.7K        0.02%           1                                     [random.c:298 -> random.c:298]          libc-2.27.so
           2.52%          198.3K        0.02%           1                                     [random.c:293 -> random.c:293]          libc-2.27.so
           2.36%          184.8K        0.02%           1                                           [rand.c:28 -> rand.c:28]          libc-2.27.so
           2.33%          180.5K        0.02%           1                                     [random.c:295 -> random.c:295]          libc-2.27.so
           2.28%          176.7K        0.02%           1                                     [random.c:295 -> random.c:295]          libc-2.27.so
           2.20%          168.8K        0.02%           1                                         [rand@plt+0 -> rand@plt+0]                   div
           1.98%          158.2K        0.02%           1                                 [random_r.c:388 -> random_r.c:388]          libc-2.27.so
           1.57%          123.3K        0.02%           1                                             [div.c:42 -> div.c:44]                   div
           1.44%          116.0K        0.42%          19                                 [random_r.c:357 -> random_r.c:394]          libc-2.27.so

--------------------------------------------------

 v7:
 ---
 1. Since we have used use_browser in report__browse_block_hists
    to support stdio mode, now we also add supporting for tui.

 2. Move block tui browser code from ui/browsers/hists.c
    to block-info.c.

 v6:
 ---
 Create report__tui_browse_block_hists in block-info.c
 (codes are moved from builtin-report.c).

 v5:
 ---
 Fix a crash issue when running perf report without
 '--total-cycles'. The issue is because the internal flag
 is renamed from 'total_cycles' to 'total_cycles_mode' in
 previous patch but this patch still uses 'total_cycles'
 to check if the '--total-cycles' option is enabled, which
 causes the code to be inconsistent.

 v4:
 ---
 Since the block collection is moved out of printing in
 previous patch, this patch is updated accordingly for
 tui supporting.

 v3:
 ---
 Minor change since the function name is changed:
 block_total_cycles_percent -> block_info__total_cycles_percent

Signed-off-by: Jin Yao <yao.jin@linux.intel.com>
Reviewed-by: Jiri Olsa <jolsa@kernel.org>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Jin Yao <yao.jin@intel.com>
Cc: Kan Liang <kan.liang@linux.intel.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lore.kernel.org/lkml/20191107074719.26139-8-yao.jin@linux.intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf report: Support --percent-limit for --total-cycles

We have already supported the '--total-cycles' option in previous patch.
It's also useful to show entries only above a threshold percent.

This patch enables '--percent-limit' for not showing entries
under that percent.

For example:

 perf report --total-cycles --stdio --percent-limit 1

 # To display the perf.data header info, please use --header/--header-only options.
 #
 #
 # Total Lost Samples: 0
 #
 # Samples: 2M of event 'cycles'
 # Event count (approx.): 2753248
 #
 # Sampled Cycles%  Sampled Cycles  Avg Cycles%  Avg Cycles                                              [Program Block Range]         Shared Object
 # ...............  ..............  ...........  ..........  .................................................................  ....................
 #
            26.04%            2.8M        0.40%          18                                             [div.c:42 -> div.c:39]                   div
            15.17%            1.2M        0.16%           7                                 [random_r.c:357 -> random_r.c:380]          libc-2.27.so
             5.11%          402.0K        0.04%           2                                             [div.c:27 -> div.c:28]                   div
             4.87%          381.6K        0.04%           2                                     [random.c:288 -> random.c:291]          libc-2.27.so
             4.53%          381.0K        0.04%           2                                             [div.c:40 -> div.c:40]                   div
             3.85%          300.9K        0.02%           1                                             [div.c:22 -> div.c:25]                   div
             3.08%          241.1K        0.02%           1                                           [rand.c:26 -> rand.c:27]          libc-2.27.so
             3.06%          240.0K        0.02%           1                                     [random.c:291 -> random.c:291]          libc-2.27.so
             2.78%          215.7K        0.02%           1                                     [random.c:298 -> random.c:298]          libc-2.27.so
             2.52%          198.3K        0.02%           1                                     [random.c:293 -> random.c:293]          libc-2.27.so
             2.36%          184.8K        0.02%           1                                           [rand.c:28 -> rand.c:28]          libc-2.27.so
             2.33%          180.5K        0.02%           1                                     [random.c:295 -> random.c:295]          libc-2.27.so
             2.28%          176.7K        0.02%           1                                     [random.c:295 -> random.c:295]          libc-2.27.so
             2.20%          168.8K        0.02%           1                                         [rand@plt+0 -> rand@plt+0]                   div
             1.98%          158.2K        0.02%           1                                 [random_r.c:388 -> random_r.c:388]          libc-2.27.so
             1.57%          123.3K        0.02%           1                                             [div.c:42 -> div.c:44]                   div
             1.44%          116.0K        0.42%          19                                 [random_r.c:357 -> random_r.c:394]          libc-2.27.so

Committer testing:

From second exapmple onwards slightly edited for brevity:

  # perf report --total-cycles --percent-limit 2 --stdio
  # To display the perf.data header info, please use --header/--header-only options.
  #
  #
  # Total Lost Samples: 0
  #
  # Samples: 6M of event 'cycles'
  # Event count (approx.): 6299936
  #
  # Sampled Cycles%  Sampled Cycles  Avg Cycles%  Avg Cycles                                                   [Program Block Range]         Shared Object
  # ...............  ..............  ...........  ..........  ......................................................................  ....................
  #
              2.17%            1.7M        0.08%         607                                        [compiler.h:199 -> common.c:221]      [kernel.vmlinux]
  #
  # (Tip: Create an archive with symtabs to analyse on other machine: perf archive)
  #
  # perf report --total-cycles --percent-limit 1 --stdio
  # Sampled Cycles%  Sampled Cycles  Avg Cycles%  Avg Cycles                                                   [Program Block Range]         Shared Object
              2.17%            1.7M        0.08%         607                                        [compiler.h:199 -> common.c:221]      [kernel.vmlinux]
              1.75%            1.3M        8.34%       65.5K    [memset-vec-unaligned-erms.S:147 -> memset-vec-unaligned-erms.S:151]          libc-2.29.so
  #
  # perf report --total-cycles --percent-limit 0.7 --stdio
  # Sampled Cycles%  Sampled Cycles  Avg Cycles%  Avg Cycles                                                   [Program Block Range]         Shared Object
              2.17%            1.7M        0.08%         607                                        [compiler.h:199 -> common.c:221]      [kernel.vmlinux]
              1.75%            1.3M        8.34%       65.5K    [memset-vec-unaligned-erms.S:147 -> memset-vec-unaligned-erms.S:151]          libc-2.29.so
              0.72%          544.5K        0.03%         230                                      [entry_64.S:657 -> entry_64.S:662]      [kernel.vmlinux]
  #

-------------------------------------------

It only shows the entries which 'Sampled Cycles%' > 1%.

 v7:
 ---
 No functional change. Only fix the conflict issue because
 previous patches are changed.

 v6:
 ---
 No functional change. Only fix the conflict issue because
 previous patches are changed.

 v5:
 ---
 No functional change. Only fix the conflict issue because
 previous patches are changed.

 v4:
 ---
 No functional change. Only fix the build issue because
 previous patches are changed.

Signed-off-by: Jin Yao <yao.jin@linux.intel.com>
Reviewed-by: Jiri Olsa <jolsa@kernel.org>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Jin Yao <yao.jin@intel.com>
Cc: Kan Liang <kan.liang@linux.intel.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lore.kernel.org/lkml/20191107074719.26139-7-yao.jin@linux.intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf report: Sort by sampled cycles percent per block for stdio

It would be useful to support sorting for all blocks by the sampled
cycles percent per block. This is useful to concentrate on the globally
hottest blocks.

This patch implements a new option "--total-cycles" which sorts all
blocks by 'Sampled Cycles%'. The 'Sampled Cycles%' is the percent:

 percent = block sampled cycles aggregation / total sampled cycles

Note that, this patch only supports "--stdio" mode.

For example,

  # perf record -b ./div
  # perf report --total-cycles --stdio
  # To display the perf.data header info, please use --header/--header-only options.
  #
  # Total Lost Samples: 0
  #
  # Samples: 2M of event 'cycles'
  # Event count (approx.): 2753248
  #
  # Sampled Cycles%  Sampled Cycles  Avg Cycles%  Avg Cycles                             [Program Block Range]      Shared Object
  # ...............  ..............  ...........  ..........  ................................................  .................
  #
             26.04%            2.8M        0.40%          18                            [div.c:42 -> div.c:39]                div
             15.17%            1.2M        0.16%           7                [random_r.c:357 -> random_r.c:380]       libc-2.27.so
              5.11%          402.0K        0.04%           2                            [div.c:27 -> div.c:28]                div
              4.87%          381.6K        0.04%           2                    [random.c:288 -> random.c:291]       libc-2.27.so
              4.53%          381.0K        0.04%           2                            [div.c:40 -> div.c:40]                div
              3.85%          300.9K        0.02%           1                            [div.c:22 -> div.c:25]                div
              3.08%          241.1K        0.02%           1                          [rand.c:26 -> rand.c:27]       libc-2.27.so
              3.06%          240.0K        0.02%           1                    [random.c:291 -> random.c:291]       libc-2.27.so
              2.78%          215.7K        0.02%           1                    [random.c:298 -> random.c:298]       libc-2.27.so
              2.52%          198.3K        0.02%           1                    [random.c:293 -> random.c:293]       libc-2.27.so
              2.36%          184.8K        0.02%           1                          [rand.c:28 -> rand.c:28]       libc-2.27.so
              2.33%          180.5K        0.02%           1                    [random.c:295 -> random.c:295]       libc-2.27.so
              2.28%          176.7K        0.02%           1                    [random.c:295 -> random.c:295]       libc-2.27.so
              2.20%          168.8K        0.02%           1                        [rand@plt+0 -> rand@plt+0]                div
              1.98%          158.2K        0.02%           1                [random_r.c:388 -> random_r.c:388]       libc-2.27.so
              1.57%          123.3K        0.02%           1                            [div.c:42 -> div.c:44]                div
              1.44%          116.0K        0.42%          19                [random_r.c:357 -> random_r.c:394]       libc-2.27.so
              0.25%          182.5K        0.02%           1                [random_r.c:388 -> random_r.c:391]       libc-2.27.so
              0.00%              48        1.07%          48        [x86_pmu_enable+284 -> x86_pmu_enable+298]  [kernel.kallsyms]
              0.00%              74        1.64%          74             [vm_mmap_pgoff+0 -> vm_mmap_pgoff+92]  [kernel.kallsyms]
              0.00%              73        1.62%          73                         [vm_mmap+0 -> vm_mmap+48]  [kernel.kallsyms]
              0.00%              63        0.69%          31                       [up_write+0 -> up_write+34]  [kernel.kallsyms]
              0.00%              13        0.29%          13      [setup_arg_pages+396 -> setup_arg_pages+413]  [kernel.kallsyms]
              0.00%               3        0.07%           3      [setup_arg_pages+418 -> setup_arg_pages+450]  [kernel.kallsyms]
              0.00%             616        6.84%         308   [security_mmap_file+0 -> security_mmap_file+72]  [kernel.kallsyms]
              0.00%              23        0.51%          23  [security_mmap_file+77 -> security_mmap_file+87]  [kernel.kallsyms]
              0.00%               4        0.02%           1                  [sched_clock+0 -> sched_clock+4]  [kernel.kallsyms]
              0.00%               4        0.02%           1                 [sched_clock+9 -> sched_clock+12]  [kernel.kallsyms]
              0.00%               1        0.02%           1                [rcu_nmi_exit+0 -> rcu_nmi_exit+9]  [kernel.kallsyms]

Committer testing:

This should provide material for hours of endless joy, both from looking
for suspicious things in the implementation of this patch, such as the
top one:

  # Sampled Cycles%  Sampled Cycles  Avg Cycles%  Avg Cycles                          [Program Block Range]     Shared Object
              2.17%            1.7M        0.08%         607   [compiler.h:199 -> common.c:221]              [kernel.vmlinux]

As well from things that look legit:

  # Sampled Cycles%  Sampled Cycles  Avg Cycles%  Avg Cycles                          [Program Block Range]     Shared Object
              0.16%          123.0K        0.60%        4.7K   [nospec-branch.h:265 -> nospec-branch.h:278]  [kernel.vmlinux]

:-)

Very short system wide taken branches session:

  # perf record -h -b

   Usage: perf record [<options>] [<command>]
      or: perf record [<options>] -- <command> [<options>]

      -b, --branch-any      sample any taken branches

  #
  # perf record -b
  ^C[ perf record: Woken up 595 times to write data ]
  [ perf record: Captured and wrote 156.672 MB perf.data (196873 samples) ]

  #
  # perf evlist -v
  cycles: size: 112, { sample_period, sample_freq }: 4000, sample_type: IP|TID|TIME|CPU|PERIOD|BRANCH_STACK, read_format: ID, disabled: 1, inherit: 1, mmap: 1, comm: 1, freq: 1, task: 1, precise_ip: 3, sample_id_all: 1, exclude_guest: 1, mmap2: 1, comm_exec: 1, ksymbol: 1, bpf_event: 1, branch_sample_type: ANY
  #
  # perf report --total-cycles --stdio
  # To display the perf.data header info, please use --header/--header-only options.
  #
  # Total Lost Samples: 0
  #
  # Samples: 6M of event 'cycles'
  # Event count (approx.): 6299936
  #
  # Sampled Cycles%  Sampled Cycles  Avg Cycles%  Avg Cycles                                                   [Program Block Range]         Shared Object
  # ...............  ..............  ...........  ..........  ......................................................................  ....................
  #
              2.17%            1.7M        0.08%         607                                        [compiler.h:199 -> common.c:221]      [kernel.vmlinux]
              1.75%            1.3M        8.34%       65.5K    [memset-vec-unaligned-erms.S:147 -> memset-vec-unaligned-erms.S:151]          libc-2.29.so
              0.72%          544.5K        0.03%         230                                      [entry_64.S:657 -> entry_64.S:662]      [kernel.vmlinux]
              0.56%          541.8K        0.09%         672                                        [compiler.h:199 -> common.c:300]      [kernel.vmlinux]
              0.39%          293.2K        0.01%         104                                    [list_debug.c:43 -> list_debug.c:61]      [kernel.vmlinux]
              0.36%          278.6K        0.03%         272                                    [entry_64.S:1289 -> entry_64.S:1308]      [kernel.vmlinux]
              0.30%          260.8K        0.07%         564                              [clear_page_64.S:47 -> clear_page_64.S:50]      [kernel.vmlinux]
              0.28%          215.3K        0.05%         369                                            [traps.c:623 -> traps.c:628]      [kernel.vmlinux]
              0.23%          178.1K        0.04%         278                                      [entry_64.S:271 -> entry_64.S:275]      [kernel.vmlinux]
              0.20%          152.6K        0.09%         706                                      [paravirt.c:177 -> paravirt.c:179]      [kernel.vmlinux]
              0.20%          155.8K        0.05%         373                                      [entry_64.S:153 -> entry_64.S:175]      [kernel.vmlinux]
              0.18%          136.6K        0.03%         222                                                [msr.h:105 -> msr.h:166]      [kernel.vmlinux]
              0.16%          123.0K        0.60%        4.7K                            [nospec-branch.h:265 -> nospec-branch.h:278]      [kernel.vmlinux]
              0.16%          118.3K        0.01%          44                                      [entry_64.S:632 -> entry_64.S:657]      [kernel.vmlinux]
              0.14%          104.5K        0.00%          28                                          [rwsem.c:1541 -> rwsem.c:1544]      [kernel.vmlinux]
              0.13%           99.2K        0.01%          53                                      [spinlock.c:150 -> spinlock.c:152]      [kernel.vmlinux]
              0.13%           95.5K        0.00%          35                                              [swap.c:456 -> swap.c:471]      [kernel.vmlinux]
              0.12%           96.2K        0.05%         407                              [copy_user_64.S:175 -> copy_user_64.S:209]      [kernel.vmlinux]
              0.11%           85.9K        0.00%          31                                        [swap.c:400 -> page-flags.h:188]      [kernel.vmlinux]
              0.10%           73.0K        0.01%          52                                          [paravirt.h:763 -> list.h:131]      [kernel.vmlinux]
              0.07%           56.2K        0.03%         214                                      [filemap.c:1524 -> filemap.c:1557]      [kernel.vmlinux]
              0.07%           54.2K        0.02%         145                                        [memory.c:1032 -> memory.c:1049]      [kernel.vmlinux]
              0.07%           50.3K        0.00%          39                                            [mmzone.c:49 -> mmzone.c:69]      [kernel.vmlinux]
              0.06%           48.3K        0.01%          40                                   [paravirt.h:768 -> page_alloc.c:3304]      [kernel.vmlinux]
              0.06%           46.7K        0.02%         155                                        [memory.c:1032 -> memory.c:1056]      [kernel.vmlinux]
              0.06%           46.9K        0.01%         103                                              [swap.c:867 -> swap.c:902]      [kernel.vmlinux]
              0.06%           47.8K        0.00%          34                                    [entry_64.S:1201 -> entry_64.S:1202]      [kernel.vmlinux]

 -----------------------------------------------------------

 v7:
 ---
 Use use_browser in report__browse_block_hists for supporting
 stdio and potential tui mode.

 v6:
 ---
 Create report__browse_block_hists in block-info.c (codes are
 moved from builtin-report.c). It's called from
 perf_evlist__tty_browse_hists.

 v5:
 ---
 1. Move all block functions to block-info.c

 2. Move the code of setting ms in block hist_entry to
    other patch.

 v4:
 ---
 1. Use new option '--total-cycles' to replace
    '-s total_cycles' in v3.

 2. Move block info collection out of block info
    printing.

 v3:
 ---
 1. Use common function block_info__process_sym to
    process the blocks per symbol.

 2. Remove the nasty hack for skipping calculation
    of column length

 3. Some minor cleanup

Signed-off-by: Jin Yao <yao.jin@linux.intel.com>
Reviewed-by: Jiri Olsa <jolsa@kernel.org>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Jin Yao <yao.jin@intel.com>
Cc: Kan Liang <kan.liang@linux.intel.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lore.kernel.org/lkml/20191107074719.26139-6-yao.jin@linux.intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>