The ASUS ROG X16 (GV601R) series laptop has the same node-to-DAC pairs
as early models and the G14, this includes bass speakers which are by
default mapped incorrectly to the 0x06 node.
The initial fix for ASUS G533Z was based on faulty information. This
fixes the pincfg to values that have been verified with no existing
module options or other hacks enabled.
Enables headphone jack, and 5.1 surround.
[ corrected the indent level by tiwai ]
Fixes: a585977433e7 ("ALSA: hda/realtek: Add pincfg for ASUS G533Z HP jack") Signed-off-by: Luke D. Jones <luke@ljones.dev> Cc: <stable@vger.kernel.org> Link: https://lore.kernel.org/r/20221010065702.35190-1-luke@ljones.dev Signed-off-by: Takashi Iwai <tiwai@suse.de> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
After some feedback from users with Dell Precision 5530 machines, this
patch reverts the previous change to add ALC289_FIXUP_DUAL_SPK.
While it improved the speaker output quality, it caused the headphone
jack to have an audible "pop" sound when power saving was toggled.
At an error path to release URB buffers and contexts, the driver might
hit a NULL dererence for u->urb pointer, when u->buffer_size has been
already set but the actual URB allocation failed.
Fix it by adding the NULL check of urb. Also, make sure that
buffer_size is cleared after the error path or the close.
When the driver hits -ENOMEM at allocating a URB or a buffer, it
aborts and goes to the error path that releases the all previously
allocated resources. However, when -ENOMEM hits at the middle of the
sync EP URB allocation loop, the partially allocated URBs might be
left without released, because ep->nurbs is still zero at that point.
Fix it by setting ep->nurbs at first, so that the error handler loops
over the full URB list.
The register_mutex taken around the dev_unregister callback call in
snd_rawmidi_free() may potentially lead to a mutex deadlock, when OSS
emulation and a hot unplug are involved.
Since the mutex doesn't protect the actual race (as the registration
itself is already protected by another means), let's drop it.
We took sound_oss_mutex around the calls of unregister_sound_special()
at unregistering OSS devices. This may, however, lead to a deadlock,
because we manage the card release via the card's device object, and
the release may happen at unregister_sound_special() call -- which
will take sound_oss_mutex again in turn.
Although the deadlock might be fixed by relaxing the rawmidi mutex in
the previous commit, it's safer to move unregister_sound_special()
calls themselves out of the sound_oss_mutex, too. The call is
race-safe as the function has a spinlock protection by itself.
Johannes Berg [Fri, 14 Oct 2022 16:47:05 +0000 (18:47 +0200)]
wifi: mac80211: fix MBSSID parsing use-after-free
Commit f279124b905209953805d9fbf77cb4277766ae17 upstream.
This is a different version of the commit, changed to store
the non-transmitted profile in the elems, and freeing it in
the few places where it's relevant, since that is only the
case when the last argument for parsing (the non-tx BSSID)
is non-NULL.
When we parse a multi-BSSID element, we might point some
element pointers into the allocated nontransmitted_profile.
However, we free this before returning, causing UAF when the
relevant pointers in the parsed elements are accessed.
Fix this by not allocating the scratch buffer separately but
as part of the returned structure instead, that way, there
are no lifetime issues with it.
The scratch buffer introduction as part of the returned data
here is taken from MLO feature work done by Ilan.
This fixes CVE-2022-42719.
Fixes: abd48b8066f3 ("mac80211: support profile split between elements") Co-developed-by: Ilan Peer <ilan.peer@intel.com> Signed-off-by: Ilan Peer <ilan.peer@intel.com> Reviewed-by: Kees Cook <keescook@chromium.org> Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Johannes Berg [Fri, 14 Oct 2022 16:47:03 +0000 (18:47 +0200)]
mac80211: mlme: find auth challenge directly
There's no need to parse all elements etc. just to find the
authentication challenge - use cfg80211_find_elem() instead.
This also allows us to remove WLAN_EID_CHALLENGE handling
from the element parsing entirely.
Suspending and resuming the system can sometimes cause the out
URB to get hung after a reset_resume. This causes LED setting
and force feedback to break on resume. To avoid this, just drop
the reset_resume callback so the USB core rebinds xpad to the
wireless pads on resume if a reset happened.
A nice side effect of this change is the LED ring on wireless
controllers is now set correctly on system resume.
When updating beacon elements in a non-transmitted BSS,
also update the hidden sub-entries to the same beacon
elements, so that a future update through other paths
won't trigger a WARN_ON().
The warning is triggered because the beacon elements in
the hidden BSSes that are children of the BSS should
always be the same as in the parent.
Reported-by: Sönke Huster <shuster@seemoo.tu-darmstadt.de> Tested-by: Sönke Huster <shuster@seemoo.tu-darmstadt.de> Fixes: ac2a324087a0 ("cfg80211: Parsing of Multiple BSSID information in scanning") Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
If the tool on the other side (e.g. wmediumd) gets confused
about the rate, we hit a warning in mac80211. Silence that
by effectively duplicating the check here and dropping the
frame silently (in mac80211 it's dropped with the warning).
If a non-transmitted BSS shares enough information (both
SSID and BSSID!) with another non-transmitted BSS of a
different AP, then we can find and update it, and then
try to add it to the non-transmitted BSS list. We do a
search for it on the transmitted BSS, but if it's not
there (but belongs to another transmitted BSS), the list
gets corrupted.
Since this is an erroneous situation, simply fail the
list insertion in this case and free the non-transmitted
BSS.
This fixes CVE-2022-42721.
Reported-by: Sönke Huster <shuster@seemoo.tu-darmstadt.de> Tested-by: Sönke Huster <shuster@seemoo.tu-darmstadt.de> Fixes: ac2a324087a0 ("cfg80211: Parsing of Multiple BSSID information in scanning") Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
There are multiple refcounting bugs related to multi-BSSID:
- In bss_ref_get(), if the BSS has a hidden_beacon_bss, then
the bss pointer is overwritten before checking for the
transmitted BSS, which is clearly wrong. Fix this by using
the bss_from_pub() macro.
- In cfg80211_bss_update() we copy the transmitted_bss pointer
from tmp into new, but then if we release new, we'll unref
it erroneously. We already set the pointer and ref it, but
need to NULL it since it was copied from the tmp data.
- In cfg80211_inform_single_bss_data(), if adding to the non-
transmitted list fails, we unlink the BSS and yet still we
return it, but this results in returning an entry without
a reference. We shouldn't return it anyway if it was broken
enough to not get added there.
Per spec, the maximum value for the MaxBSSID ('n') indicator is 8,
and the minimum is 1 since a multiple BSSID set with just one BSSID
doesn't make sense (the # of BSSIDs is limited by 2^n).
Limit this in the parsing in both cfg80211 and mac80211, rejecting
any elements with an invalid value.
This fixes potentially bad shifts in the processing of these inside
the cfg80211_gen_new_bssid() function later.
I found this during the investigation of CVE-2022-41674 fixed by the
previous patch.
Fixes: ac2a324087a0 ("cfg80211: Parsing of Multiple BSSID information in scanning") Fixes: 7d7dcb365227 ("mac80211: support multi-bssid") Reviewed-by: Kees Cook <keescook@chromium.org> Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In the copy code of the elements, we do the following calculation
to reach the end of the MBSSID element:
/* copy the IEs after MBSSID */
cpy_len = mbssid[1] + 2;
This looks fine, however, cpy_len is a u8, the same as mbssid[1],
so the addition of two can overflow. In this case the subsequent
memcpy() will overflow the allocated buffer, since it copies 256
bytes too much due to the way the allocation and memcpy() sizes
are calculated.
Fix this by using size_t for the cpy_len variable.
This fixes CVE-2022-41674.
Reported-by: Soenke Huster <shuster@seemoo.tu-darmstadt.de> Tested-by: Soenke Huster <shuster@seemoo.tu-darmstadt.de> Fixes: ac2a324087a0 ("cfg80211: Parsing of Multiple BSSID information in scanning") Reviewed-by: Kees Cook <keescook@chromium.org> Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Previously, the fast pool was dumped into the main pool periodically in
the fast pool's hard IRQ handler. This worked fine and there weren't
problems with it, until RT came around. Since RT converts spinlocks into
sleeping locks, problems cropped up. Rather than switching to raw
spinlocks, the RT developers preferred we make the transformation from
originally doing:
This is an ordinary pattern done all over the kernel. However, Sherry
noticed a 10% performance regression in qperf TCP over a 40gbps
InfiniBand card. Quoting her message:
> MT27500 Family [ConnectX-3] cards:
> Infiniband device 'mlx4_0' port 1 status:
> default gid: fe80:0000:0000:0000:0010:e000:0178:9eb1
> base lid: 0x6
> sm lid: 0x1
> state: 4: ACTIVE
> phys state: 5: LinkUp
> rate: 40 Gb/sec (4X QDR)
> link_layer: InfiniBand
>
> Cards are configured with IP addresses on private subnet for IPoIB
> performance testing.
> Regression identified in this bug is in TCP latency in this stack as reported
> by qperf tcp_lat metric:
>
> We have one system listen as a qperf server:
> [root@yourQperfServer ~]# qperf
>
> Have the other system connect to qperf server as a client (in this
> case, it’s X7 server with Mellanox card):
> [root@yourQperfClient ~]# numactl -m0 -N0 qperf 20.20.20.101 -v -uu -ub --time 60 --wait_server 20 -oo msg_size:4K:1024K:*2 tcp_lat
Rather than incur the scheduling latency from queue_work_on, we can
instead switch to running on the next timer tick, on the same core. This
also batches things a bit more -- once per jiffy -- which is okay now
that mix_interrupt_randomness() can credit multiple bits at once.
Reported-by: Sherry Yang <sherry.yang@oracle.com> Tested-by: Paul Webb <paul.x.webb@oracle.com> Cc: Sherry Yang <sherry.yang@oracle.com> Cc: Phillip Goerl <phillip.goerl@oracle.com> Cc: Jack Vogel <jack.vogel@oracle.com> Cc: Nicky Veitch <nicky.veitch@oracle.com> Cc: Colm Harrington <colm.harrington@oracle.com> Cc: Ramanan Govindarajan <ramanan.govindarajan@oracle.com> Cc: Sebastian Andrzej Siewior <bigeasy@linutronix.de> Cc: Dominik Brodowski <linux@dominikbrodowski.net> Cc: Tejun Heo <tj@kernel.org> Cc: Sultan Alsawaf <sultan@kerneltoast.com> Cc: stable@vger.kernel.org Fixes: 59647cfb0716 ("random: defer fast pool mixing to worker") Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In order to avoid reading and dirtying two cache lines on every IRQ,
move the work_struct to the bottom of the fast_pool struct. add_
interrupt_randomness() always touches .pool and .count, which are
currently split, because .mix pushes everything down. Instead, move .mix
to the bottom, so that .pool and .count are always in the first cache
line, since .mix is only accessed when the pool is full.
Fixes: 59647cfb0716 ("random: defer fast pool mixing to worker") Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Prior to 5.6, when /dev/random was opened with O_NONBLOCK, it would
return -EAGAIN if there was no entropy. When the pools were unified in
5.6, this was lost. The post 5.6 behavior of blocking until the pool is
initialized, and ignoring O_NONBLOCK in the process, went unnoticed,
with no reports about the regression received for two and a half years.
However, eventually this indeed did break somebody's userspace.
So we restore the old behavior, by returning -EAGAIN if the pool is not
initialized. Unlike the old /dev/random, this can only occur during
early boot, after which it never blocks again.
In order to make this O_NONBLOCK behavior consistent with other
expectations, also respect users reading with preadv2(RWF_NOWAIT) and
similar.
Fixes: 76176a3b7001 ("random: make /dev/random be almost like /dev/urandom") Reported-by: Guozihua <guozihua@huawei.com> Reported-by: Zhongguohua <zhongguohua1@huawei.com> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Theodore Ts'o <tytso@mit.edu> Cc: Andrew Lutomirski <luto@kernel.org> Cc: stable@vger.kernel.org Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The passthrough structure is declared off of the stack, so it needs to be
set to zero before copied back to userspace to prevent any unintentional
data leakage. Switch things to be statically allocated which will fill the
unused fields with 0 automatically.
Link: https://lore.kernel.org/r/YxrjN3OOw2HHl9tx@kroah.com Cc: stable@kernel.org Cc: "James E.J. Bottomley" <jejb@linux.ibm.com> Cc: "Martin K. Petersen" <martin.petersen@oracle.com> Cc: Dan Carpenter <dan.carpenter@oracle.com> Reported-by: hdthky <hdthky0@gmail.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Hans reported that his Sony VAIO VPX11S1E showed the broken sound
behavior at the start of the stream for a couple of seconds, and it
turned out that the position_fix=1 option fixes the issue. It implies
that the position reporting is inaccurate, and very likely hitting on
all Poulsbo devices.
The patch applies the workaround for Poulsbo generically to switch to
LPIB mode instead of the default position buffer.
Since the most that's mixed into the pool is sizeof(long)*2, don't
credit more than that many bytes of entropy.
Fixes: bcd9fd8727b6 ("random: do not use input pool from hard IRQs") Cc: stable@vger.kernel.org Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
If creation or finalization of a checkpoint fails due to anomalies in the
checkpoint metadata on disk, a kernel warning is generated.
This patch replaces the WARN_ONs by nilfs_error, so that a kernel, booted
with panic_on_warn, does not panic. A nilfs_error is appropriate here to
handle the abnormal filesystem condition.
This also replaces the detected error codes with an I/O error so that
neither of the internal error codes is returned to callers.
If nilfs_attach_log_writer() failed to create a log writer thread, it
frees a data structure of the log writer without any cleanup. After
commit 693b9b4ce61c ("nilfs2: use root object to get ifile"), this causes
a leak of struct nilfs_root, which started to leak an ifile metadata inode
and a kobject on that struct.
In addition, if the kernel is booted with panic_on_warn, the above
ifile metadata inode leak will cause the following panic when the
nilfs2 kernel module is removed:
If the i_mode field in inode of metadata files is corrupted on disk, it
can cause the initialization of bmap structure, which should have been
called from nilfs_read_inode_common(), not to be called. This causes a
lockdep warning followed by a NULL pointer dereference at
nilfs_bmap_lookup_at_level().
This patch fixes these issues by adding a missing sanitiy check for the
i_mode field of metadata file's inode.
The use of strncpy() is considered deprecated for NUL-terminated
strings[1]. Replace strncpy() with strscpy_pad(), to keep existing
pad-behavior of strncpy, similarly to commit ee37b1929c67 ("rpmsg:
glink: Replace strncpy() with strscpy_pad()"). This fixes W=1 warning:
In function ‘qcom_glink_rx_close’,
inlined from ‘qcom_glink_work’ at ../drivers/rpmsg/qcom_glink_native.c:1638:4:
drivers/rpmsg/qcom_glink_native.c:1549:17: warning: ‘strncpy’ specified bound 32 equals destination size [-Wstringop-truncation]
1549 | strncpy(chinfo.name, channel->name, sizeof(chinfo.name));
This loop intends to retry a max of 10 times, with some implicit
termination based on the SD_{R,}OCR_S18A bit. Unfortunately, the
termination condition depends on the value reported by the SD card
(*rocr), which may or may not correctly reflect what we asked it to do.
Needless to say, it's not wise to rely on the card doing what we expect;
we should at least terminate the loop regardless. So, check both the
input and output values, so we ensure we will terminate regardless of
the SD card behavior.
Note that SDIO learned a similar retry loop in commit ba45bf740a48
("mmc: core: Fixup signal voltage switch"), but that used the 'ocr'
result, and so the current pre-terminating condition looks like:
rocr & ocr & R4_18V_PRESENT
(i.e., it doesn't have the same bug.)
This addresses a number of crash reports seen on ChromeOS that look
like the following:
... // lots of repeated: ...
<4>[13142.846061] mmc1: Skipping voltage switch
<4>[13143.406087] mmc1: Skipping voltage switch
<4>[13143.964724] mmc1: Skipping voltage switch
<4>[13144.526089] mmc1: Skipping voltage switch
<4>[13145.086088] mmc1: Skipping voltage switch
<4>[13145.645941] mmc1: Skipping voltage switch
<3>[13146.153969] INFO: task halt:30352 blocked for more than 122 seconds.
...
Fixes: 0cc861603562 ("mmc: sd: add support for signal voltage switch procedure") Cc: <stable@vger.kernel.org> Signed-off-by: Brian Norris <briannorris@chromium.org> Reviewed-by: Guenter Roeck <linux@roeck-us.net> Link: https://lore.kernel.org/r/20220914014010.2076169-1-briannorris@chromium.org Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Syzbot found an issue in usbmon module, where the user space client can
corrupt the monitor's internal memory, causing the usbmon module to
crash the kernel with segfault, UAF, etc.
The reproducer mmaps the /dev/usbmon memory to user space, and
overwrites it with arbitrary data, which causes all kinds of issues.
Return an -EPERM error from mon_bin_mmap() if the flag VM_WRTIE is set.
Also clear VM_MAYWRITE to make it impossible to change it to writable
later.
Since binutils 2.39, ld will print a warning if any stack section is
executable, which is the default for stack sections on files without a
.note.GNU-stack section.
This was fixed for x86 in commit 60ef90f725e4 ("x86: link vdso and boot with -z noexecstack --no-warn-rwx-segments"),
but remained broken for UML, resulting in several warnings:
/usr/bin/ld: warning: arch/x86/um/vdso/vdso.o: missing .note.GNU-stack section implies executable stack
/usr/bin/ld: NOTE: This behaviour is deprecated and will be removed in a future version of the linker
/usr/bin/ld: warning: .tmp_vmlinux.kallsyms1 has a LOAD segment with RWX permissions
/usr/bin/ld: warning: .tmp_vmlinux.kallsyms1.o: missing .note.GNU-stack section implies executable stack
/usr/bin/ld: NOTE: This behaviour is deprecated and will be removed in a future version of the linker
/usr/bin/ld: warning: .tmp_vmlinux.kallsyms2 has a LOAD segment with RWX permissions
/usr/bin/ld: warning: .tmp_vmlinux.kallsyms2.o: missing .note.GNU-stack section implies executable stack
/usr/bin/ld: NOTE: This behaviour is deprecated and will be removed in a future version of the linker
/usr/bin/ld: warning: vmlinux has a LOAD segment with RWX permissions
Link both the VDSO and vmlinux with -z noexecstack, fixing the warnings
about .note.GNU-stack sections. In addition, pass --no-warn-rwx-segments
to dodge the remaining warnings about LOAD segments with RWX permissions
in the kallsyms objects. (Note that this flag is apparently not
available on lld, so hide it behind a test for BFD, which is what the
x86 patch does.)
arch.tls_array is statically allocated so checking for NULL doesn't
make sense. This causes the compiler warning below.
Remove the checks to silence these warnings.
../arch/x86/um/tls_32.c: In function 'get_free_idx':
../arch/x86/um/tls_32.c:68:13: warning: the comparison will always evaluate as 'true' for the address of 'tls_array' will never be NULL [-Waddress]
68 | if (!t->arch.tls_array)
| ^
In file included from ../arch/x86/um/asm/processor.h:10,
from ../include/linux/rcupdate.h:30,
from ../include/linux/rculist.h:11,
from ../include/linux/pid.h:5,
from ../include/linux/sched.h:14,
from ../arch/x86/um/tls_32.c:7:
../arch/x86/um/asm/processor_32.h:22:31: note: 'tls_array' declared here
22 | struct uml_tls_struct tls_array[GDT_ENTRY_TLS_ENTRIES];
| ^~~~~~~~~
../arch/x86/um/tls_32.c: In function 'get_tls_entry':
../arch/x86/um/tls_32.c:243:13: warning: the comparison will always evaluate as 'true' for the address of 'tls_array' will never be NULL [-Waddress]
243 | if (!t->arch.tls_array)
| ^
../arch/x86/um/asm/processor_32.h:22:31: note: 'tls_array' declared here
22 | struct uml_tls_struct tls_array[GDT_ENTRY_TLS_ENTRIES];
| ^~~~~~~~~
Signed-off-by: Lukas Straub <lukasstraub2@web.de> Acked-by: Randy Dunlap <rdunlap@infradead.org> # build-tested Signed-off-by: Richard Weinberger <richard@nod.at> Signed-off-by: Sasha Levin <sashal@kernel.org>
There is uninit value bug in dgram_sendmsg function in
net/ieee802154/socket.c when the length of valid data pointed by the
msg->msg_name isn't verified.
We introducing a helper function ieee802154_sockaddr_check_size to
check namelen. First we check there is addr_type in ieee802154_addr_sa.
Then, we check namelen according to addr_type.
Also fixed in raw_bind, dgram_bind, dgram_connect.
Signed-off-by: Haimin Zhang <tcs_kernel@tencent.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Sasha Levin <sashal@kernel.org>
In __qedf_probe(), if qedf->cdev is NULL which means
qed_ops->common->probe() failed, then the program will goto label err1, and
scsi_host_put() will free lport->host pointer. Because the memory qedf
points to is allocated by libfc_host_alloc(), it will be freed by
scsi_host_put(). However, the if statement below label err0 only checks
whether qedf is NULL but doesn't check whether the memory has been freed.
So a UAF bug can occur.
There are two ways to reach the statements below err0. The first one is
described as before, "qedf" should be set to NULL. The second one is goto
"err0" directly. In the latter scenario qedf hasn't been changed and it has
the initial value NULL. As a result the if statement is not reachable in
any situation.
In alloc_inode, inode_init_always() could return -ENOMEM if
security_inode_alloc() fails, which causes inode->i_private
uninitialized. Then nilfs_is_metadata_file_inode() returns
true and nilfs_free_inode() wrongly calls nilfs_mdt_destroy(),
which frees the uninitialized inode->i_private
and leads to crashes(e.g., UAF/GPF).
Fix this by moving security_inode_alloc just prior to
this_cpu_inc(nr_inodes)
Link: https://lkml.kernel.org/r/CAFcO6XOcf1Jj2SeGt=jJV59wmhESeSKpfR0omdFRq+J9nD1vfQ@mail.gmail.com Reported-by: butt3rflyh4ck <butterflyhuangxx@gmail.com> Reported-by: Hao Sun <sunhao.th@gmail.com> Reported-by: Jiacheng Xu <stitch@zju.edu.cn> Reviewed-by: Christian Brauner (Microsoft) <brauner@kernel.org> Signed-off-by: Dongliang Mu <mudongliangabcd@gmail.com> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: stable@vger.kernel.org Signed-off-by: Al Viro <viro@zeniv.linux.org.uk> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The mmap lock protects the page walker from changes to the page tables
during the walk. However a read lock is insufficient to protect those
areas which don't have a VMA as munmap() detaches the VMAs before
downgrading to a read lock and actually tearing down PTEs/page tables.
For users of walk_page_range() the solution is to simply call pte_hole()
immediately without checking the actual page tables when a VMA is not
present. We now never call __walk_page_range() without a valid vma.
For walk_page_range_novma() the locking requirements are tightened to
require the mmap write lock to be taken, and then walking the pgd
directly with 'no_vma' set.
This in turn means that all page walkers either have a valid vma, or
it's that special 'novma' case for page table debugging. As a result,
all the odd '(!walk->vma && !walk->no_vma)' tests can be removed.
Fixes: 5dcd5782f466 ("mm: mmap: zap pages with read mmap_sem in munmap") Reported-by: Jann Horn <jannh@google.com> Signed-off-by: Steven Price <steven.price@arm.com> Cc: Vlastimil Babka <vbabka@suse.cz> Cc: Thomas Hellström <thomas.hellstrom@linux.intel.com> Cc: Konstantin Khlebnikov <koct9i@gmail.com> Cc: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
[manually backported. backport note: walk_page_range_novma() does not exist in
5.4, so I'm omitting it from the backport] Signed-off-by: Jann Horn <jannh@google.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
We enable -Wcast-function-type globally in the kernel to warn about
mismatching types in function pointer casts. Compilers currently
warn only about ABI incompability with this flag, but Clang 16 will
enable a stricter version of the check by default that checks for an
exact type match. This will be very noisy in the kernel, so disable
-Wcast-function-type-strict without W=1 until the new warnings have
been addressed.
fs/xfs/xfs_inode.c: In function 'xfs_itruncate_extents_flags':
fs/xfs/xfs_inode.c:1523:8: warning: unused variable 'done' [-Wunused-variable]
commit 65eb0bb1eeff ("xfs: truncate should remove
all blocks, not just to the end of the page cache")
left behind this, so remove it.
Fixes: 65eb0bb1eeff ("xfs: truncate should remove all blocks, not just to the end of the page cache") Reported-by: Hulk Robot <hulkci@huawei.com> Reported-by: Stephen Rothwell <sfr@canb.auug.org.au> Signed-off-by: YueHaibing <yuehaibing@huawei.com> Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com> Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com> Acked-by: Darrick J. Wong <djwong@kernel.org> Signed-off-by: Chandan Babu R <chandan.babu@oracle.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Dan Carpenter pointed out that error is uninitialized. While there
never should be an attr leaf block with zero entries, let's not leave
that logic bomb there.
Fixes: 92ee175e92e8 ("xfs: streamline xfs_attr3_leaf_inactive") Reported-by: Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com> Reviewed-by: Allison Collins <allison.henderson@oracle.com> Reviewed-by: Eric Sandeen <sandeen@redhat.com> Acked-by: Darrick J. Wong <djwong@kernel.org> Signed-off-by: Chandan Babu R <chandan.babu@oracle.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Now that we know we don't have to take a transaction to stale the incore
buffers for a remote value, get rid of the unnecessary memory allocation
in the leaf walker and call the rmt_stale function directly. Flatten
the loop while we're at it.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Acked-by: Darrick J. Wong <djwong@kernel.org> Signed-off-by: Chandan Babu R <chandan.babu@oracle.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Move the abstract in-memory version of various btree block headers
out of xfs_da_format.h as they aren't on-disk formats.
Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com> Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com> Acked-by: Darrick J. Wong <djwong@kernel.org> Signed-off-by: Chandan Babu R <chandan.babu@oracle.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[Replaced XFS_IS_CORRUPT() calls with ASSERT() for 5.4.y backport]
While running generic/103, I observed what looks like memory corruption
and (with slub debugging turned on) a slub redzone warning on i386 when
inactivating an inode with a 64k remote attr value.
On a v5 filesystem, maximally sized remote attr values require one block
more than 64k worth of space to hold both the remote attribute value
header (64 bytes). On a 4k block filesystem this results in a 68k
buffer; on a 64k block filesystem, this would be a 128k buffer. Note
that even though we'll never use more than 65,600 bytes of this buffer,
XFS_MAX_BLOCKSIZE is 64k.
This is a problem because the definition of struct xfs_buf_log_format
allows for XFS_MAX_BLOCKSIZE worth of dirty bitmap (64k). On i386 when we
invalidate a remote attribute, xfs_trans_binval zeroes all 68k worth of
the dirty map, writing right off the end of the log item and corrupting
memory. We've gotten away with this on x86_64 for years because the
compiler inserts a u32 padding on the end of struct xfs_buf_log_format.
Fortunately for us, remote attribute values are written to disk with
xfs_bwrite(), which is to say that they are not logged. Fix the problem
by removing all places where we could end up creating a buffer log item
for a remote attribute value and leave a note explaining why. Next,
replace the open-coded buffer invalidation with a call to the helper we
created in the previous patch that does better checking for bad metadata
before marking the buffer stale.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Acked-by: Darrick J. Wong <djwong@kernel.org> Signed-off-by: Chandan Babu R <chandan.babu@oracle.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[Replaced XFS_IS_CORRUPT() calls with ASSERT() for 5.4.y backport]
Hoist the code that invalidates remote extended attribute value buffers
into a separate helper function. This prepares us for a memory
corruption fix in the next patch.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Acked-by: Darrick J. Wong <djwong@kernel.org> Signed-off-by: Chandan Babu R <chandan.babu@oracle.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Direct I/O reads can also be used with RWF_NOWAIT & co. Fix the inode
locking in xfs_file_dio_aio_read to take IOCB_NOWAIT into account.
Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Carlos Maiolino <cmaiolino@redhat.com> Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com> Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com> Acked-by: Darrick J. Wong <djwong@kernel.org> Signed-off-by: Chandan Babu R <chandan.babu@oracle.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
I observed a hang in generic/308 while running fstests on a i686 kernel.
The hang occurred when trying to purge the pagecache on a large sparse
file that had a page created past MAX_LFS_FILESIZE, which caused an
integer overflow in the pagecache xarray and resulted in an infinite
loop.
I then noticed that Linus changed the definition of MAX_LFS_FILESIZE in
commit 43dd2082cc51 ("Clarify (and fix) MAX_LFS_FILESIZE macros") so
that it is now one page short of the maximum page index on 32-bit
kernels. Because the XFS function to compute max offset open-codes the
2005-era MAX_LFS_FILESIZE computation and neither the vfs nor mm perform
any sanity checking of s_maxbytes, the code in generic/308 can create a
page above the pagecache's limit and kaboom.
Fix all this by setting s_maxbytes to MAX_LFS_FILESIZE directly and
aborting the mount with a warning if our assumptions ever break. I have
no answer for why this seems to have been broken for years and nobody
noticed.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Acked-by: Darrick J. Wong <djwong@kernel.org> Signed-off-by: Chandan Babu R <chandan.babu@oracle.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
xfs_itruncate_extents_flags() is supposed to unmap every block in a file
from EOF onwards. Oddly, it uses s_maxbytes as the upper limit to the
bunmapi range, even though s_maxbytes reflects the highest offset the
pagecache can support, not the highest offset that XFS supports.
The result of this confusion is that if you create a 20T file on a
64-bit machine, mount the filesystem on a 32-bit machine, and remove the
file, we leak everything above 16T. Fix this by capping the bunmapi
request at the maximum possible block offset, not s_maxbytes.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Acked-by: Darrick J. Wong <djwong@kernel.org> Signed-off-by: Chandan Babu R <chandan.babu@oracle.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Introduce a new #define for the maximum supported file block offset.
We'll use this in the next patch to make it more obvious that we're
doing some operation for all possible inode fork mappings after a given
offset. We can't use ULLONG_MAX here because bunmapi uses that to
detect when it's done.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Acked-by: Darrick J. Wong <djwong@kernel.org> Signed-off-by: Chandan Babu R <chandan.babu@oracle.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
XFS_ATTR_INCOMPLETE is a flag in the on-disk attribute format, and thus
in a different namespace as the ATTR_* flags in xfs_da_args.flags.
Switch to using a XFS_DA_OP_INCOMPLETE flag in op_flags instead. Without
this users might be able to inject this flag into operations using the
attr by handle ioctl.
Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com> Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com> Acked-by: Darrick J. Wong <djwong@kernel.org> Signed-off-by: Chandan Babu R <chandan.babu@oracle.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
tl;dr: The Enhanced IBRS mitigation for Spectre v2 does not work as
documented for RET instructions after VM exits. Mitigate it with a new
one-entry RSB stuffing mechanism and a new LFENCE.
== Background ==
Indirect Branch Restricted Speculation (IBRS) was designed to help
mitigate Branch Target Injection and Speculative Store Bypass, i.e.
Spectre, attacks. IBRS prevents software run in less privileged modes
from affecting branch prediction in more privileged modes. IBRS requires
the MSR to be written on every privilege level change.
To overcome some of the performance issues of IBRS, Enhanced IBRS was
introduced. eIBRS is an "always on" IBRS, in other words, just turn
it on once instead of writing the MSR on every privilege level change.
When eIBRS is enabled, more privileged modes should be protected from
less privileged modes, including protecting VMMs from guests.
== Problem ==
Here's a simplification of how guests are run on Linux' KVM:
void run_kvm_guest(void)
{
// Prepare to run guest
VMRESUME();
// Clean up after guest runs
}
The execution flow for that would look something like this to the
processor:
1. Host-side: call run_kvm_guest()
2. Host-side: VMRESUME
3. Guest runs, does "CALL guest_function"
4. VM exit, host runs again
5. Host might make some "cleanup" function calls
6. Host-side: RET from run_kvm_guest()
Now, when back on the host, there are a couple of possible scenarios of
post-guest activity the host needs to do before executing host code:
* on pre-eIBRS hardware (legacy IBRS, or nothing at all), the RSB is not
touched and Linux has to do a 32-entry stuffing.
* on eIBRS hardware, VM exit with IBRS enabled, or restoring the host
IBRS=1 shortly after VM exit, has a documented side effect of flushing
the RSB except in this PBRSB situation where the software needs to stuff
the last RSB entry "by hand".
IOW, with eIBRS supported, host RET instructions should no longer be
influenced by guest behavior after the host retires a single CALL
instruction.
However, if the RET instructions are "unbalanced" with CALLs after a VM
exit as is the RET in #6, it might speculatively use the address for the
instruction after the CALL in #3 as an RSB prediction. This is a problem
since the (untrusted) guest controls this address.
Balanced CALL/RET instruction pairs such as in step #5 are not affected.
== Solution ==
The PBRSB issue affects a wide variety of Intel processors which
support eIBRS. But not all of them need mitigation. Today,
X86_FEATURE_RSB_VMEXIT triggers an RSB filling sequence that mitigates
PBRSB. Systems setting RSB_VMEXIT need no further mitigation - i.e.,
eIBRS systems which enable legacy IBRS explicitly.
However, such systems (X86_FEATURE_IBRS_ENHANCED) do not set RSB_VMEXIT
and most of them need a new mitigation.
Therefore, introduce a new feature flag X86_FEATURE_RSB_VMEXIT_LITE
which triggers a lighter-weight PBRSB mitigation versus RSB_VMEXIT.
The lighter-weight mitigation performs a CALL instruction which is
immediately followed by a speculative execution barrier (INT3). This
steers speculative execution to the barrier -- just like a retpoline
-- which ensures that speculation can never reach an unbalanced RET.
Then, ensure this CALL is retired before continuing execution with an
LFENCE.
In other words, the window of exposure is opened at VM exit where RET
behavior is troublesome. While the window is open, force RSB predictions
sampling for RET targets to a dead end at the INT3. Close the window
with the LFENCE.
There is a subset of eIBRS systems which are not vulnerable to PBRSB.
Add these systems to the cpu_vuln_whitelist[] as NO_EIBRS_PBRSB.
Future systems that aren't vulnerable will set ARCH_CAP_PBRSB_NO.
[ bp: Massage, incorporate review comments from Andy Cooper. ]
Signed-off-by: Daniel Sneddon <daniel.sneddon@linux.intel.com> Co-developed-by: Pawan Gupta <pawan.kumar.gupta@linux.intel.com> Signed-off-by: Pawan Gupta <pawan.kumar.gupta@linux.intel.com> Signed-off-by: Borislav Petkov <bp@suse.de>
[cascardo: no intra-function validation] Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@canonical.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
IBRS mitigation for spectre_v2 forces write to MSR_IA32_SPEC_CTRL at
every kernel entry/exit. On Enhanced IBRS parts setting
MSR_IA32_SPEC_CTRL[IBRS] only once at boot is sufficient. MSR writes at
every kernel entry/exit incur unnecessary performance loss.
When Enhanced IBRS feature is present, print a warning about this
unnecessary performance loss.
Signed-off-by: Pawan Gupta <pawan.kumar.gupta@linux.intel.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Thadeu Lima de Souza Cascardo <cascardo@canonical.com> Cc: stable@vger.kernel.org Link: https://lore.kernel.org/r/2a5eaf54583c2bfe0edc4fea64006656256cca17.1657814857.git.pawan.kumar.gupta@linux.intel.com Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@canonical.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Some Intel processors may use alternate predictors for RETs on
RSB-underflow. This condition may be vulnerable to Branch History
Injection (BHI) and intramode-BTI.
Kernel earlier added spectre_v2 mitigation modes (eIBRS+Retpolines,
eIBRS+LFENCE, Retpolines) which protect indirect CALLs and JMPs against
such attacks. However, on RSB-underflow, RET target prediction may
fallback to alternate predictors. As a result, RET's predicted target
may get influenced by branch history.
A new MSR_IA32_SPEC_CTRL bit (RRSBA_DIS_S) controls this fallback
behavior when in kernel mode. When set, RETs will not take predictions
from alternate predictors, hence mitigating RETs as well. Support for
this is enumerated by CPUID.7.2.EDX[RRSBA_CTRL] (bit2).
For spectre v2 mitigation, when a user selects a mitigation that
protects indirect CALLs and JMPs against BHI and intramode-BTI, set
RRSBA_DIS_S also to protect RETs for RSB-underflow case.
Signed-off-by: Pawan Gupta <pawan.kumar.gupta@linux.intel.com> Signed-off-by: Borislav Petkov <bp@suse.de>
[cascardo: no tools/arch/x86/include/asm/msr-index.h] Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@canonical.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
BTC_NO indicates that hardware is not susceptible to Branch Type Confusion.
Zen3 CPUs don't suffer BTC.
Hypervisors are expected to synthesise BTC_NO when it is appropriate
given the migration pool, to prevent kernels using heuristics.
[ bp: Massage. ]
Signed-off-by: Andrew Cooper <andrew.cooper3@citrix.com> Signed-off-by: Borislav Petkov <bp@suse.de> Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@canonical.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Prevent RSB underflow/poisoning attacks with RSB. While at it, add a
bunch of comments to attempt to document the current state of tribal
knowledge about RSB attacks and what exactly is being mitigated.
Signed-off-by: Josh Poimboeuf <jpoimboe@kernel.org> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Signed-off-by: Borislav Petkov <bp@suse.de> Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@canonical.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Convert __vmx_vcpu_run()'s 'launched' argument to 'flags', in
preparation for doing SPEC_CTRL handling immediately after vmexit, which
will need another flag.
This is much easier than adding a fourth argument, because this code
supports both 32-bit and 64-bit, and the fourth argument on 32-bit would
have to be pushed on the stack.
Note that __vmx_vcpu_run_flags() is called outside of the noinstr
critical section because it will soon start calling potentially
traceable functions.
Signed-off-by: Josh Poimboeuf <jpoimboe@kernel.org> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Signed-off-by: Borislav Petkov <bp@suse.de> Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@canonical.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Replace inline assembly in nested_vmx_check_vmentry_hw
with a call to __vmx_vcpu_run. The function is not
performance critical, so (double) GPR save/restore
in __vmx_vcpu_run can be tolerated, as far as performance
effects are concerned.
Cc: Paolo Bonzini <pbonzini@redhat.com> Cc: Sean Christopherson <seanjc@google.com> Reviewed-and-tested-by: Sean Christopherson <seanjc@google.com> Signed-off-by: Uros Bizjak <ubizjak@gmail.com>
[sean: dropped versioning info from changelog] Signed-off-by: Sean Christopherson <seanjc@google.com>
Message-Id: <20201231002702.2223707-5-seanjc@google.com> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
[cascardo: small fixups] Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@canonical.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This mask has been made redundant by kvm_spec_ctrl_test_value(). And it
doesn't even work when MSR interception is disabled, as the guest can
just write to SPEC_CTRL directly.
Signed-off-by: Josh Poimboeuf <jpoimboe@kernel.org> Signed-off-by: Borislav Petkov <bp@suse.de> Reviewed-by: Paolo Bonzini <pbonzini@redhat.com> Signed-off-by: Borislav Petkov <bp@suse.de> Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@canonical.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
If a kernel is built with CONFIG_RETPOLINE=n, but the user still wants
to mitigate Spectre v2 using IBRS or eIBRS, the RSB filling will be
silently disabled.
There's nothing retpoline-specific about RSB buffer filling. Remove the
CONFIG_RETPOLINE guards around it.
Signed-off-by: Josh Poimboeuf <jpoimboe@kernel.org> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Signed-off-by: Borislav Petkov <bp@suse.de> Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@canonical.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Change FILL_RETURN_BUFFER so that objtool groks it and can generate
correct ORC unwind information.
- Since ORC is alternative invariant; that is, all alternatives
should have the same ORC entries, the __FILL_RETURN_BUFFER body
can not be part of an alternative.
Therefore, move it out of the alternative and keep the alternative
as a sort of jump_label around it.
- Use the ANNOTATE_INTRA_FUNCTION_CALL annotation to white-list
these 'funny' call instructions to nowhere.
- Use UNWIND_HINT_EMPTY to 'fill' the speculation traps, otherwise
objtool will consider them unreachable.
- Move the RSP adjustment into the loop, such that the loop has a
deterministic stack layout.
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Reviewed-by: Alexandre Chartre <alexandre.chartre@oracle.com> Acked-by: Josh Poimboeuf <jpoimboe@redhat.com> Link: https://lkml.kernel.org/r/20200428191700.032079304@infradead.org
[cascardo: fixup because of backport of 35b035a983dc6e8f1f427a38a8bb8b6c945f421d ("x86/speculation: Add LFENCE to RSB fill sequence")]
[cascardo: no intra-function call validation support]
[cascardo: avoid UNWIND_HINT_EMPTY because of svm] Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@canonical.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Having IBRS enabled while the SMT sibling is idle unnecessarily slows
down the running sibling. OTOH, disabling IBRS around idle takes two
MSR writes, which will increase the idle latency.
Therefore, only disable IBRS around deeper idle states. Shallow idle
states are bounded by the tick in duration, since NOHZ is not allowed
for them by virtue of their short target residency.
Only do this for mwait-driven idle, since that keeps interrupts disabled
across idle, which makes disabling IBRS vs IRQ-entry a non-issue.
Note: C6 is a random threshold, most importantly C1 probably shouldn't
disable IBRS, benchmarking needed.
Suggested-by: Tim Chen <tim.c.chen@linux.intel.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Signed-off-by: Borislav Petkov <bp@suse.de> Reviewed-by: Josh Poimboeuf <jpoimboe@kernel.org> Signed-off-by: Borislav Petkov <bp@suse.de>
[cascardo: no CPUIDLE_FLAG_IRQ_ENABLE] Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@canonical.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[cascardo: context adjustments] Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@canonical.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Implement Kernel IBRS - currently the only known option to mitigate RSB
underflow speculation issues on Skylake hardware.
Note: since IBRS_ENTER requires fuller context established than
UNTRAIN_RET, it must be placed after it. However, since UNTRAIN_RET
itself implies a RET, it must come after IBRS_ENTER. This means
IBRS_ENTER needs to also move UNTRAIN_RET.
Note 2: KERNEL_IBRS is sub-optimal for XenPV.
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Signed-off-by: Borislav Petkov <bp@suse.de> Reviewed-by: Josh Poimboeuf <jpoimboe@kernel.org> Signed-off-by: Borislav Petkov <bp@suse.de>
[cascardo: conflict at arch/x86/entry/entry_64.S, skip_r11rcx]
[cascardo: conflict at arch/x86/entry/entry_64_compat.S]
[cascardo: conflict fixups, no ANNOTATE_NOENDBR]
[cascardo: entry fixups because of missing UNTRAIN_RET]
[cascardo: conflicts on fsgsbase] Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@canonical.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Yes, r11 and rcx have been restored previously, but since they're being
popped anyway (into rsi) might as well pop them into their own regs --
setting them to the value they already are.
Less magical code.
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Signed-off-by: Borislav Petkov <bp@suse.de> Link: https://lore.kernel.org/r/20220506121631.365070674@infradead.org Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@canonical.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Due to TIF_SSBD and TIF_SPEC_IB the actual IA32_SPEC_CTRL value can
differ from x86_spec_ctrl_base. As such, keep a per-CPU value
reflecting the current task's MSR content.
[jpoimboe: rename]
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Signed-off-by: Borislav Petkov <bp@suse.de> Reviewed-by: Josh Poimboeuf <jpoimboe@kernel.org> Signed-off-by: Borislav Petkov <bp@suse.de> Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@canonical.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Add the "retbleed=<value>" boot parameter to select a mitigation for
RETBleed. Possible values are "off", "auto" and "unret"
(JMP2RET mitigation). The default value is "auto".
Currently, "retbleed=auto" will select the unret mitigation on
AMD and Hygon and no mitigation on Intel (JMP2RET is not effective on
Intel).
[peterz: rebase; add hygon]
[jpoimboe: cleanups]
Signed-off-by: Alexandre Chartre <alexandre.chartre@oracle.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Signed-off-by: Borislav Petkov <bp@suse.de> Reviewed-by: Josh Poimboeuf <jpoimboe@kernel.org> Signed-off-by: Borislav Petkov <bp@suse.de>
[cascardo: this effectively remove the UNRET mitigation as an option, so it
has to be complemented by a later pick of the same commit later. This is
done in order to pick retbleed_select_mitigation] Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@canonical.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Intel uses the same family/model for several CPUs. Sometimes the
stepping must be checked to tell them apart.
On x86 there can be at most 16 steppings. Add a steppings bitmask to
x86_cpu_id and a X86_MATCH_VENDOR_FAMILY_MODEL_STEPPING_FEATURE macro
and support for matching against family/model/stepping.
[ bp: Massage. ]
Signed-off-by: Mark Gross <mgross@linux.intel.com> Signed-off-by: Borislav Petkov <bp@suse.de> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Reviewed-by: Tony Luck <tony.luck@intel.com> Reviewed-by: Josh Poimboeuf <jpoimboe@redhat.com>
[cascardo: have steppings be the last member as there are initializers
that don't use named members] Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@canonical.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
projects / kernel.git / log