Am 03.03.21 um 02:17 schrieb Anthony DeRossi:
Freed pages are not subtracted from the allocated_pages counter in
ttm_pool_type_fini(), causing a leak in the count on device removal.
The next shrinker invocation loops forever trying to free pages that are
no longer in the pool:
rcu: INFO: rcu_sched self-detected stall on CPU
rcu: 3-....: (9998 ticks this GP) idle=54e/1/0x4000000000000000 softirq=434857/434857 fqs=2237
(t=10001 jiffies g=2194533 q=49211)
NMI backtrace for cpu 3
CPU: 3 PID: 1034 Comm: kswapd0 Tainted: P O 5.11.0-com #1
Hardware name: System manufacturer System Product Name/PRIME X570-PRO, BIOS 1405 11/19/2019
Call Trace:
<IRQ>
...
</IRQ>
sysvec_apic_timer_interrupt+0x77/0x80
asm_sysvec_apic_timer_interrupt+0x12/0x20
RIP: 0010:mutex_unlock+0x16/0x20
Code: e7 48 8b 70 10 e8 7a 53 77 ff eb aa e8 43 6c ff ff 0f 1f 00 65 48 8b 14 25 00 6d 01 00 31 c9 48 89 d0 f0 48 0f b1 0f 48 39 c2 <74> 05 e9 e3 fe ff ff c3 66 90 48 8b 47 20 48 85 c0 74 0f 8b 50 10
RSP: 0018:ffffbdb840797be8 EFLAGS: 00000246
RAX: ffff9ff445a41c00 RBX: ffffffffc02a9ef8 RCX: 0000000000000000
RDX: ffff9ff445a41c00 RSI: ffffbdb840797c78 RDI: ffffffffc02a9ac0
RBP: 0000000000000080 R08: 0000000000000000 R09: ffffbdb840797c80
R10: 0000000000000000 R11: fffffffffffffff5 R12: 0000000000000000
R13: 0000000000000000 R14: 0000000000000084 R15: ffffffffc02a9a60
ttm_pool_shrink+0x7d/0x90 [ttm]
ttm_pool_shrinker_scan+0x5/0x20 [ttm]
do_shrink_slab+0x13a/0x1a0
shrink_slab+0xcb/0x2f0
shrink_node+0x2c8/0x6c0
balance_pgdat+0x2f7/0x580
kswapd+0x17c/0x300
? wait_woken+0x70/0x70
? balance_pgdat+0x580/0x580
kthread+0xff/0x140
? kthread_park+0x80/0x80
ret_from_fork+0x22/0x30
debugfs shows the incorrect total:
$ cat /sys/kernel/debug/dri/0/ttm_page_pool
--- 0--- --- 1--- --- 2--- --- 3--- --- 4--- --- 5--- --- 6--- --- 7--- --- 8--- --- 9--- ---10---
wc : 0 0 0 0 0 0 0 0 0 0 0
uc : 0 0 0 0 0 0 0 0 0 0 0
wc 32 : 0 0 0 0 0 0 0 0 0 0 0
uc 32 : 0 0 0 0 0 0 0 0 0 0 0
DMA uc : 0 0 0 0 0 0 0 0 0 0 0
DMA wc : 0 0 0 0 0 0 0 0 0 0 0
DMA : 0 0 0 0 0 0 0 0 0 0 0
total : 3029 of 8244261
Using ttm_pool_type_take() to remove pages from the pool before freeing
them correctly accounts for the freed pages.
Fixes: d099fc8f540a ("drm/ttm: new TT backend allocation pool v3")
Signed-off-by: Anthony DeRossi <ajderossi@xxxxxxxxx>
Good catch. Patch is Reviewed-by: Christian König <christian.koenig@xxxxxxx>
Going to push that to drm-misc-fixes for upstreaming in a minute.
Thanks,
Christian.
---
drivers/gpu/drm/ttm/ttm_pool.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/drivers/gpu/drm/ttm/ttm_pool.c b/drivers/gpu/drm/ttm/ttm_pool.c
index 74bf1c84b637..f6df79034805 100644
--- a/drivers/gpu/drm/ttm/ttm_pool.c
+++ b/drivers/gpu/drm/ttm/ttm_pool.c
@@ -258,13 +258,13 @@ static void ttm_pool_type_init(struct ttm_pool_type *pt, struct ttm_pool *pool,
/* Remove a pool_type from the global shrinker list and free all pages */
static void ttm_pool_type_fini(struct ttm_pool_type *pt)
{
- struct page *p, *tmp;
+ struct page *p;
mutex_lock(&shrinker_lock);
list_del(&pt->shrinker_list);
mutex_unlock(&shrinker_lock);
- list_for_each_entry_safe(p, tmp, &pt->pages, lru)
+ while ((p = ttm_pool_type_take(pt)))
ttm_pool_free_page(pt->pool, pt->caching, pt->order, p);
}
_______________________________________________
dri-devel mailing list
dri-devel@xxxxxxxxxxxxxxxxxxxxx
https://lists.freedesktop.org/mailman/listinfo/dri-devel
