On 6/26/22 6:04 PM, Jens Axboe wrote: > On 6/26/22 4:56 PM, Greg Thelen wrote: >> Greg Kroah-Hartman <gregkh@xxxxxxxxxxxxxxxxxxx> wrote: >> >>> I'm announcing the release of the 5.10.125 kernel. >>> >>> All users of the 5.10 kernel series must upgrade. >>> >>> The updated 5.10.y git tree can be found at: >>> git://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable.git linux-5.10.y >>> and can be browsed at the normal kernel.org git web browser: >>> https://git.kernel.org/?p=linux/kernel/git/stable/linux-stable.git;a=summary >>> >>> thanks, >>> >>> greg k-h >>> >>> ------------ >>> >>> Makefile | 2 >>> arch/arm64/mm/cache.S | 2 >>> arch/s390/mm/pgtable.c | 2 >>> drivers/tty/serial/serial_core.c | 34 ++++-------- >>> drivers/usb/gadget/function/u_ether.c | 11 +++- >>> fs/io_uring.c | 23 +++++--- >>> fs/zonefs/super.c | 92 ++++++++++++++++++++++------------ >>> net/ipv4/inet_hashtables.c | 31 ++++++++--- >>> 8 files changed, 122 insertions(+), 75 deletions(-) >>> >>> Christian Borntraeger (1): >>> s390/mm: use non-quiescing sske for KVM switch to keyed guest >>> >>> Damien Le Moal (1): >>> zonefs: fix zonefs_iomap_begin() for reads >>> >>> Eric Dumazet (1): >>> tcp: add some entropy in __inet_hash_connect() >>> >>> Greg Kroah-Hartman (1): >>> Linux 5.10.125 >>> >>> Jens Axboe (1): >>> io_uring: add missing item types for various requests >>> >>> Lukas Wunner (1): >>> serial: core: Initialize rs485 RTS polarity already on probe >>> >>> Marian Postevca (1): >>> usb: gadget: u_ether: fix regression in setting fixed MAC address >>> >>> Will Deacon (1): >>> arm64: mm: Don't invalidate FROM_DEVICE buffers at start of DMA transfer >>> >>> Willy Tarreau (5): >>> tcp: use different parts of the port_offset for index and offset >>> tcp: add small random increments to the source port >>> tcp: dynamically allocate the perturb table used by source ports >>> tcp: increase source port perturb table to 2^16 >>> tcp: drop the hash_32() part from the index calculation >> >> 5.10.125 commit df3f3bb5059d20ef094d6b2f0256c4bf4127a859 ("io_uring: add >> missing item types for various requests") causes panic when running >> test/iopoll.t from https://github.com/axboe/liburing commit >> dda4848a9911120a903bef6284fb88286f4464c9 (liburing-2.2). >> >> Here's a manually annotated panic message: >> [ 359.047161] list_del corruption, ffffa42098824f80->next is LIST_POISON1 (dead000000000100) >> [ 359.055393] kernel BUG at lib/list_debug.c:47! >> [ 359.059786] invalid opcode: 0000 [#1] SMP DEBUG_PAGEALLOC PTI >> [ 359.065463] CPU: 11 PID: 15862 Comm: iopoll.t Tainted: G S I 5.10.124 #1 >> [ 359.081804] RIP: 0010:__list_del_entry_valid+0x49/0x80 >> [ 359.086880] Code: c2 22 48 39 d1 74 25 48 8b 11 48 39 f2 75 2d 48 8b 50 08 48 39 f2 75 34 b0 01 5d c3 48 c7 c7 68 15 79 b1 31 c0 e8 c5 a2 5a 00 <0f> 0b 48 c7 c7 d8 8e 76 b1 31 c0 e8 b5 a2 5a 00 0f 0b 48 c7 c7 69 >> [ 359.105431] RSP: 0018:ffffb6b66785bd58 EFLAGS: 00010046 >> [ 359.110592] RAX: 000000000000004e RBX: ffffa42098824f00 RCX: d07284ea1fbba400 >> [ 359.117642] RDX: ffffa43f7f4f05b8 RSI: ffffa43f7f4dff48 RDI: ffffa43f7f4dff48 >> [ 359.124691] RBP: ffffb6b66785bd58 R08: 0000000000000000 R09: ffffffffb1f38540 >> [ 359.131740] R10: 00000000ffff7fff R11: 0000000000000000 R12: 0000000000000282 >> [ 359.138789] R13: ffffb6b66785beb8 R14: ffffa42095d33e00 R15: ffffa420937e3d20 >> [ 359.145836] FS: 00000000004f8380(0000) GS:ffffa43f7f4c0000(0000) knlGS:0000000000000000 >> [ 359.153830] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 >> [ 359.159506] CR2: 0000000000539388 CR3: 000000027b57c006 CR4: 00000000003706e0 >> [ 359.166552] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 >> [ 359.173600] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400 >> [ 359.180647] Call Trace: >> [ 359.183064] io_dismantle_req+0x1da/0x2b0 >> __list_del_entry [include/linux/list.h:132] >> list_del [include/linux/list.h:146] >> io_req_drop_files [fs/io_uring.c:5934] >> io_req_clean_work [fs/io_uring.c:1315] >> io_dismantle_req [fs/io_uring.c:1911] >> [ 359.187023] io_do_iopoll+0x4e5/0x790 >> [ 359.194602] __se_sys_io_uring_enter+0x39b/0x6f0 >> [ 359.208318] __x64_sys_io_uring_enter+0x29/0x30 >> [ 359.212793] do_syscall_64+0x31/0x40 >> [ 359.216324] entry_SYSCALL_64_after_hwframe+0x44/0xa9 > > Well that sucks, I wonder why mine didn't fail like that. I'll see if I > can hit this and send a fix. Thanks for reporting! Below should do it, I apologize for that. I think my test box booted the previous kernel which is why it didn't hit it in my regression tests :-( Greg, can you add this to 5.10-stable? Verified it ran tests with the right kernel now... commit d2aaf8fdc544a3edf44820a07294d693dedd12e3 Author: Jens Axboe <axboe@xxxxxxxxx> Date: Sun Jun 26 18:16:17 2022 -0600 io_uring: use separate list entry for iopoll requests A previous commit ended up enabling file tracking for iopoll requests, which conflicts with both of them using the same list entry for tracking. Add a separate list entry just for iopoll requests, avoid this issue. No upstream commit exists for this issue. Reported-by: Greg Thelen <gthelen@xxxxxxxxxx> Fixes: df3f3bb5059d ("io_uring: add missing item types for various requests") Signed-off-by: Jens Axboe <axboe@xxxxxxxxx> diff --git a/fs/io_uring.c b/fs/io_uring.c index 40ac37beca47..2e12dcbc7b0f 100644 --- a/fs/io_uring.c +++ b/fs/io_uring.c @@ -696,6 +696,8 @@ struct io_kiocb { */ struct list_head inflight_entry; + struct list_head iopoll_entry; + struct percpu_ref *fixed_file_refs; struct callback_head task_work; /* for polled requests, i.e. IORING_OP_POLL_ADD and async armed poll */ @@ -2350,8 +2352,8 @@ static void io_iopoll_queue(struct list_head *again) struct io_kiocb *req; do { - req = list_first_entry(again, struct io_kiocb, inflight_entry); - list_del(&req->inflight_entry); + req = list_first_entry(again, struct io_kiocb, iopoll_entry); + list_del(&req->iopoll_entry); __io_complete_rw(req, -EAGAIN, 0, NULL); } while (!list_empty(again)); } @@ -2373,14 +2375,14 @@ static void io_iopoll_complete(struct io_ring_ctx *ctx, unsigned int *nr_events, while (!list_empty(done)) { int cflags = 0; - req = list_first_entry(done, struct io_kiocb, inflight_entry); + req = list_first_entry(done, struct io_kiocb, iopoll_entry); if (READ_ONCE(req->result) == -EAGAIN) { req->result = 0; req->iopoll_completed = 0; - list_move_tail(&req->inflight_entry, &again); + list_move_tail(&req->iopoll_entry, &again); continue; } - list_del(&req->inflight_entry); + list_del(&req->iopoll_entry); if (req->flags & REQ_F_BUFFER_SELECTED) cflags = io_put_rw_kbuf(req); @@ -2416,7 +2418,7 @@ static int io_do_iopoll(struct io_ring_ctx *ctx, unsigned int *nr_events, spin = !ctx->poll_multi_file && *nr_events < min; ret = 0; - list_for_each_entry_safe(req, tmp, &ctx->iopoll_list, inflight_entry) { + list_for_each_entry_safe(req, tmp, &ctx->iopoll_list, iopoll_entry) { struct kiocb *kiocb = &req->rw.kiocb; /* @@ -2425,7 +2427,7 @@ static int io_do_iopoll(struct io_ring_ctx *ctx, unsigned int *nr_events, * and complete those lists first, if we have entries there. */ if (READ_ONCE(req->iopoll_completed)) { - list_move_tail(&req->inflight_entry, &done); + list_move_tail(&req->iopoll_entry, &done); continue; } if (!list_empty(&done)) @@ -2437,7 +2439,7 @@ static int io_do_iopoll(struct io_ring_ctx *ctx, unsigned int *nr_events, /* iopoll may have completed current req */ if (READ_ONCE(req->iopoll_completed)) - list_move_tail(&req->inflight_entry, &done); + list_move_tail(&req->iopoll_entry, &done); if (ret && spin) spin = false; @@ -2670,7 +2672,7 @@ static void io_iopoll_req_issued(struct io_kiocb *req) struct io_kiocb *list_req; list_req = list_first_entry(&ctx->iopoll_list, struct io_kiocb, - inflight_entry); + iopoll_entry); if (list_req->file != req->file) ctx->poll_multi_file = true; } @@ -2680,9 +2682,9 @@ static void io_iopoll_req_issued(struct io_kiocb *req) * it to the front so we find it first. */ if (READ_ONCE(req->iopoll_completed)) - list_add(&req->inflight_entry, &ctx->iopoll_list); + list_add(&req->iopoll_entry, &ctx->iopoll_list); else - list_add_tail(&req->inflight_entry, &ctx->iopoll_list); + list_add_tail(&req->iopoll_entry, &ctx->iopoll_list); if ((ctx->flags & IORING_SETUP_SQPOLL) && wq_has_sleeper(&ctx->sq_data->wait)) -- Jens Axboe
