Archivos mensuales: agosto 2024

ice: fix concurrent reset and removal of VFs

Commit c503e63200c6 ("ice: Stop processing VF messages during teardown")
introduced a driver state flag, ICE_VF_DEINIT_IN_PROGRESS, which is
intended to prevent some issues with concurrently handling messages from
VFs while tearing down the VFs.

This change was motivated by crashes caused while tearing down and
bringing up VFs in rapid succession.

It turns out that the fix actually introduces issues with the VF driver
caused because the PF no longer responds to any messages sent by the VF
during its .remove routine. This results in the VF potentially removing
its DMA memory before the PF has shut down the device queues.

Additionally, the fix doesn't actually resolve concurrency issues within
the ice driver. It is possible for a VF to initiate a reset just prior
to the ice driver removing VFs. This can result in the remove task
concurrently operating while the VF is being reset. This results in
similar memory corruption and panics purportedly fixed by that commit.

Fix this concurrency at its root by protecting both the reset and
removal flows using the existing VF cfg_lock. This ensures that we
cannot remove the VF while any outstanding critical tasks such as a
virtchnl message or a reset are occurring.

This locking change also fixes the root cause originally fixed by commit
c503e63200c6 ("ice: Stop processing VF messages during teardown"), so we
can simply revert it.

Note that I kept these two changes together because simply reverting the
original commit alone would leave the driver vulnerable to worse race
conditions.

bpf: Fix crash due to incorrect copy_map_value

When both bpf_spin_lock and bpf_timer are present in a BPF map value,
copy_map_value needs to skirt both objects when copying a value into and
out of the map. However, the current code does not set both s_off and
t_off in copy_map_value, which leads to a crash when e.g. bpf_spin_lock
is placed in map value with bpf_timer, as bpf_map_update_elem call will
be able to overwrite the other timer object.

When the issue is not fixed, an overwriting can produce the following
splat:

[root@(none) bpf]# ./test_progs -t timer_crash
[ 15.930339] bpf_testmod: loading out-of-tree module taints kernel.
[ 16.037849] ==================================================================
[ 16.038458] BUG: KASAN: user-memory-access in __pv_queued_spin_lock_slowpath+0x32b/0x520
[ 16.038944] Write of size 8 at addr 0000000000043ec0 by task test_progs/325
[ 16.039399]
[ 16.039514] CPU: 0 PID: 325 Comm: test_progs Tainted: G OE 5.16.0+ #278
[ 16.039983] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS ArchLinux 1.15.0-1 04/01/2014
[ 16.040485] Call Trace:
[ 16.040645]
[ 16.040805] dump_stack_lvl+0x59/0x73
[ 16.041069] ? __pv_queued_spin_lock_slowpath+0x32b/0x520
[ 16.041427] kasan_report.cold+0x116/0x11b
[ 16.041673] ? __pv_queued_spin_lock_slowpath+0x32b/0x520
[ 16.042040] __pv_queued_spin_lock_slowpath+0x32b/0x520
[ 16.042328] ? memcpy+0x39/0x60
[ 16.042552] ? pv_hash+0xd0/0xd0
[ 16.042785] ? lockdep_hardirqs_off+0x95/0xd0
[ 16.043079] __bpf_spin_lock_irqsave+0xdf/0xf0
[ 16.043366] ? bpf_get_current_comm+0x50/0x50
[ 16.043608] ? jhash+0x11a/0x270
[ 16.043848] bpf_timer_cancel+0x34/0xe0
[ 16.044119] bpf_prog_c4ea1c0f7449940d_sys_enter+0x7c/0x81
[ 16.044500] bpf_trampoline_6442477838_0+0x36/0x1000
[ 16.044836] __x64_sys_nanosleep+0x5/0x140
[ 16.045119] do_syscall_64+0x59/0x80
[ 16.045377] ? lock_is_held_type+0xe4/0x140
[ 16.045670] ? irqentry_exit_to_user_mode+0xa/0x40
[ 16.046001] ? mark_held_locks+0x24/0x90
[ 16.046287] ? asm_exc_page_fault+0x1e/0x30
[ 16.046569] ? asm_exc_page_fault+0x8/0x30
[ 16.046851] ? lockdep_hardirqs_on+0x7e/0x100
[ 16.047137] entry_SYSCALL_64_after_hwframe+0x44/0xae
[ 16.047405] RIP: 0033:0x7f9e4831718d
[ 16.047602] Code: b4 0c 00 0f 05 eb a9 66 0f 1f 44 00 00 f3 0f 1e fa 48 89 f8 48 89 f7 48 89 d6 48 89 ca 4d 89 c2 4d 89 c8 4c 8b 4c 24 08 0f 05 3d 01 f0 ff ff 73 01 c3 48 8b 0d b3 6c 0c 00 f7 d8 64 89 01 48
[ 16.048764] RSP: 002b:00007fff488086b8 EFLAGS: 00000206 ORIG_RAX: 0000000000000023
[ 16.049275] RAX: ffffffffffffffda RBX: 00007f9e48683740 RCX: 00007f9e4831718d
[ 16.049747] RDX: 0000000000000000 RSI: 0000000000000000 RDI: 00007fff488086d0
[ 16.050225] RBP: 00007fff488086f0 R08: 00007fff488085d7 R09: 00007f9e4cb594a0
[ 16.050648] R10: 0000000000000000 R11: 0000000000000206 R12: 00007f9e484cde30
[ 16.051124] R13: 0000000000000000 R14: 0000000000000000 R15: 0000000000000000
[ 16.051608]
[ 16.051762] ==================================================================