]> git.baikalelectronics.ru Git - kernel.git/commit
net: dsa: felix: fix min gate len calculation for tc when its first gate is closed
authorVladimir Oltean <vladimir.oltean@nxp.com>
Thu, 4 Aug 2022 20:28:17 +0000 (23:28 +0300)
committerJakub Kicinski <kuba@kernel.org>
Tue, 9 Aug 2022 03:51:56 +0000 (20:51 -0700)
commitc255be0c5f5f78d844a74c7d1cb5bb24c95c2a8a
tree48d702b81cc5bb73062f163cb768436df795aa1c
parent160f137940a683f729ae5d7fc3731dc19b37baf7
net: dsa: felix: fix min gate len calculation for tc when its first gate is closed

min_gate_len[tc] is supposed to track the shortest interval of
continuously open gates for a traffic class. For example, in the
following case:

TC 76543210

t0 00000001b 200000 ns
t1 00000010b 200000 ns

min_gate_len[0] and min_gate_len[1] should be 200000, while
min_gate_len[2-7] should be 0.

However what happens is that min_gate_len[0] is 200000, but
min_gate_len[1] ends up being 0 (despite gate_len[1] being 200000 at the
point where the logic detects the gate close event for TC 1).

The problem is that the code considers a "gate close" event whenever it
sees that there is a 0 for that TC (essentially it's level rather than
edge triggered). By doing that, any time a gate is seen as closed
without having been open prior, gate_len, which is 0, will be written
into min_gate_len. Once min_gate_len becomes 0, it's impossible for it
to track anything higher than that (the length of actually open
intervals).

To fix this, we make the writing to min_gate_len[tc] be edge-triggered,
which avoids writes for gates that are closed in consecutive intervals.
However what this does is it makes us need to special-case the
permanently closed gates at the end.

Fixes: eb99ed107125 ("net: dsa: felix: drop oversized frames with tc-taprio instead of hanging the port")
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Link: https://lore.kernel.org/r/20220804202817.1677572-1-vladimir.oltean@nxp.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
drivers/net/dsa/ocelot/felix_vsc9959.c