Fix reload stats structure exposed to the user. Change stats structure
hierarchy to have the reload action as a parent of the stat entry and
then stat entry includes value per limit. This will also help to avoid
string concatenation on iproute2 output.
Reload stats structure before this fix:
"stats": {
"reload": { "driver_reinit": 2, "fw_activate": 1, "fw_activate_no_reset": 0
}
}
After this fix:
"stats": {
"reload": { "driver_reinit": { "unspecified": 2
}, "fw_activate": { "unspecified": 1, "no_reset": 0
}
}
Devlink keeps its own reference to every reporter in a list and inits
refcount to 1 upon reporter's creation. Existing destructor waits to
free the memory indefinitely using msleep() until all references except
devlink's own are put.
Rework this mechanism by moving memory free routine to a separate
function, which is called when the last reporter reference is put.
Besides, it allows to call __devlink_health_reporter_destroy() while
locked on a reporters list mutex in symmetry to
__devlink_health_reporter_create(), which is required in follow-up
patch.
Signed-off-by: Vladyslav Tarasiuk <email address hidden>
Reviewed-by: Moshe Shemesh <email address hidden>
Reviewed-by: Jiri Pirko <email address hidden>
Signed-off-by: David S. Miller <email address hidden>
(cherry picked from commit 3c5584bf0a0493e8d232ade65f4b9c5e995f3a0c)
Signed-off-by: Bodong Wang <email address hidden>
Convert drop monitor to use the recently introduced
'devlink_trap_report' tracepoint instead of having devlink call into
drop monitor.
This is both consistent with software originated drops ('kfree_skb'
tracepoint) and also allows drop monitor to be built as a module and
still report hardware originated drops.
Signed-off-by: Ido Schimmel <email address hidden>
Reviewed-by: Jiri Pirko <email address hidden>
Signed-off-by: David S. Miller <email address hidden>
(backported from commit 8ee2267ad33e0ba021e9dd9b437f773906cd99d6)
Signed-off-by: Bodong Wang <email address hidden>
[bodong: ignore doc and drop monitor]
Devlink objects are accessible only after they were registered and
have valid devlink_*->devlink pointers.
Remove that check and simplify respective fill functions as an outcome
of such change.
Signed-off-by: Leon Romanovsky <email address hidden>
Signed-off-by: David S. Miller <email address hidden>
(backported from commit 7ca973dc9fe589dc0ab2650641f4c7a19cc49ecd)
Signed-off-by: Bodong Wang <email address hidden>
[bodong: fix conflict due to api change]
Conflicts:
net/core/devlink.c
777e34d...
by
Jakub Kicinski <email address hidden>
devlink: factor out building a snapshot notification
We'll need to send snapshot info back on the socket
which requested a snapshot to be created. Factor out
constructing a snapshot description from the broadcast
notification code.
v3: new patch
Signed-off-by: Jakub Kicinski <email address hidden>
Reviewed-by: Jiri Pirko <email address hidden>
Reviewed-by: Jacob Keller <email address hidden>
Signed-off-by: David S. Miller <email address hidden>
(cherry picked from commit dd86fec7e06ab792fe470c66a67ff42bf5d72b91)
Signed-off-by: Bodong Wang <email address hidden>
24d0d78...
by
Leon Romanovsky <email address hidden>
devlink: Allocate devlink directly in requested net namespace
There is no need in extra call indirection and check from impossible
flow where someone tries to set namespace without prior call
to devlink_alloc().
Instead of this extra logic and additional EXPORT_SYMBOL, use specialized
devlink allocation function that receives net namespace as an argument.
Such specialized API allows clear view when devlink initialized in wrong
net namespace and/or kernel users don't try to change devlink namespace
under the hood.
Reviewed-by: Jiri Pirko <email address hidden>
Signed-off-by: Leon Romanovsky <email address hidden>
Signed-off-by: Jakub Kicinski <email address hidden>
(backported from commit 26713455048eb19122b1561b471d30710177ef97)
Signed-off-by: Bodong Wang <email address hidden>
[bodong: ignore netdevsim]
Conflicts:
drivers/net/netdevsim/dev.c
19bb413...
by
Leon Romanovsky <email address hidden>
devlink: Break parameter notification sequence to be before/after unload/load driver
The change of namespaces during devlink reload calls to driver unload
before it accesses devlink parameters. The commands below causes to
use-after-free bug when trying to get flow steering mode.
* ip netns add n1
* devlink dev reload pci/0000:00:09.0 netns n1
==================================================================
BUG: KASAN: use-after-free in mlx5_devlink_fs_mode_get+0x96/0xa0 [mlx5_core]
Read of size 4 at addr ffff888009d04308 by task devlink/275
The buggy address belongs to the object at ffff888009d04300
which belongs to the cache kmalloc-128 of size 128
The buggy address is located 8 bytes inside of
128-byte region [ffff888009d04300, ffff888009d04380)
The buggy address belongs to the page:
page:0000000086a64ecc refcount:1 mapcount:0 mapping:0000000000000000 index:0xffff888009d04000 pfn:0x9d04
head:0000000086a64ecc order:1 compound_mapcount:0
flags: 0x4000000000010200(slab|head)
raw: 4000000000010200 ffffea0000203980 0000000200000002 ffff8880050428c0
raw: ffff888009d04000 000000008020001d 00000001ffffffff 0000000000000000
page dumped because: kasan: bad access detected
Memory state around the buggy address:
ffff888009d04200: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
ffff888009d04280: fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc
>ffff888009d04300: fa fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb ^
ffff888009d04380: fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc
ffff888009d04400: fa fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
==================================================================
The right solution to devlink reload is to notify about deletion of
parameters, unload driver, change net namespaces, load driver and notify
about addition of parameters.
Fixes: 070c63f20f6c ("net: devlink: allow to change namespaces during reload")
Reviewed-by: Parav Pandit <email address hidden>
Signed-off-by: Leon Romanovsky <email address hidden>
Signed-off-by: Jakub Kicinski <email address hidden>
(cherry picked from commit 05a7f4a8dff19999ca8a83a35ff4782689de7bfc)
Signed-off-by: Bodong Wang <email address hidden>
9468571...
by
Leon Romanovsky <email address hidden>
Both registered flag and devlink pointer are set at the same time
and indicate the same thing - devlink/devlink_port are ready. Instead
of checking ->registered use devlink pointer as an indication.
Signed-off-by: Leon Romanovsky <email address hidden>
Signed-off-by: David S. Miller <email address hidden>
(backported from commit d7907a2b1a3b89bea136025f885035a083525e41)
Signed-off-by: Bodong Wang <email address hidden>
[bodong: resolve conflict due to new line]
Conflicts:
net/core/devlink.c
61af07b...
by
Oleksandr Mazur <email address hidden>
Whenever query statistics is issued for trap, devlink subsystem
would also fill-in statistics 'dropped' field. This field indicates
the number of packets HW dropped and failed to report to the device driver,
and thus - to the devlink subsystem itself.
In case if device driver didn't register callback for hard drop
statistics querying, 'dropped' field will be omitted and not filled.
Signed-off-by: Oleksandr Mazur <email address hidden>
Signed-off-by: David S. Miller <email address hidden>
(cherry picked from commit ddee9dbc3d7aec1cd9fdcc671db2dd0016fd0f3d)
Signed-off-by: Bodong Wang <email address hidden>
Ready for review
for merging
into
~canonical-kernel/ubuntu/+source/linux-bluefield/+git/focal:master-next
