[pve-devel] [PATCH v2 qemu] io_uring: resubmit when result is -EAGAIN

[pve-devel] [PATCH v2 qemu] io_uring: resubmit when result is -EAGAIN

[Fabian Ebner]

Linux SCSI can throw spurious -EAGAIN in some corner cases in its
completion path, which will end up being the result in the completed
io_uring request.

Resubmitting such requests should allow block jobs to complete, even
if such spurious errors are encountered.

Signed-off-by: Fabian Ebner <f.ebner@proxmox.com>
---

Changes from v1:
    * make it a patch for pve-qemu instead of the qemu submodule
    * use v2 of the patch with a comment by Stefan Hajnoczi

 ...uring-resubmit-when-result-is-EAGAIN.patch | 49 +++++++++++++++++++
 debian/patches/series                         |  1 +
 2 files changed, 50 insertions(+)
 create mode 100644 debian/patches/extra/0002-block-io_uring-resubmit-when-result-is-EAGAIN.patch

diff --git a/debian/patches/extra/0002-block-io_uring-resubmit-when-result-is-EAGAIN.patch b/debian/patches/extra/0002-block-io_uring-resubmit-when-result-is-EAGAIN.patch
new file mode 100644
index 0000000..b5dd07a
--- /dev/null
+++ b/debian/patches/extra/0002-block-io_uring-resubmit-when-result-is-EAGAIN.patch
@@ -0,0 +1,49 @@
+From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
+From: Fabian Ebner <f.ebner@proxmox.com>
+Date: Fri, 23 Jul 2021 12:42:23 +0200
+Subject: [PATCH] block/io_uring: resubmit when result is -EAGAIN
+
+Linux SCSI can throw spurious -EAGAIN in some corner cases in its
+completion path, which will end up being the result in the completed
+io_uring request.
+
+Resubmitting such requests should allow block jobs to complete, even
+if such spurious errors are encountered.
+
+Co-authored-by: Stefan Hajnoczi <stefanha@gmail.com>
+Reviewed-by: Stefano Garzarella <sgarzare@redhat.com>
+Signed-off-by: Fabian Ebner <f.ebner@proxmox.com>
+---
+ block/io_uring.c | 16 +++++++++++++++-
+ 1 file changed, 15 insertions(+), 1 deletion(-)
+
+diff --git a/block/io_uring.c b/block/io_uring.c
+index 00a3ee9fb8..dfa475cc87 100644
+--- a/block/io_uring.c
++++ b/block/io_uring.c
+@@ -165,7 +165,21 @@ static void luring_process_completions(LuringState *s)
+         total_bytes = ret + luringcb->total_read;
+ 
+         if (ret < 0) {
+-            if (ret == -EINTR) {
++            /*
++             * Only writev/readv/fsync requests on regular files or host block
++             * devices are submitted. Therefore -EAGAIN is not expected but it's
++             * known to happen sometimes with Linux SCSI. Submit again and hope
++             * the request completes successfully.
++             *
++             * For more information, see:
++             * https://lore.kernel.org/io-uring/20210727165811.284510-3-axboe@kernel.dk/T/#u
++             *
++             * If the code is changed to submit other types of requests in the
++             * future, then this workaround may need to be extended to deal with
++             * genuine -EAGAIN results that should not be resubmitted
++             * immediately.
++             */
++            if (ret == -EINTR || ret == -EAGAIN) {
+                 luring_resubmit(s, luringcb);
+                 continue;
+             }
+-- 
+2.30.2
+
diff --git a/debian/patches/series b/debian/patches/series
index ada674b..6edf8ce 100644
--- a/debian/patches/series
+++ b/debian/patches/series
@@ -1,4 +1,5 @@
 extra/0001-monitor-qmp-fix-race-on-CHR_EVENT_CLOSED-without-OOB.patch
+extra/0002-block-io_uring-resubmit-when-result-is-EAGAIN.patch
 bitmap-mirror/0001-drive-mirror-add-support-for-sync-bitmap-mode-never.patch
 bitmap-mirror/0002-drive-mirror-add-support-for-conditional-and-always-.patch
 bitmap-mirror/0003-mirror-add-check-for-bitmap-mode-without-bitmap.patch
-- 
2.30.2

[pve-devel] applied: Re: [PATCH v2 qemu] io_uring: resubmit when result is -EAGAIN

[Thomas Lamprecht]

On 29/07/2021 11:50, Fabian Ebner wrote:
> Linux SCSI can throw spurious -EAGAIN in some corner cases in its
> completion path, which will end up being the result in the completed
> io_uring request.
> 
> Resubmitting such requests should allow block jobs to complete, even
> if such spurious errors are encountered.
> 
> Signed-off-by: Fabian Ebner <f.ebner@proxmox.com>
> ---
> 
> Changes from v1:
>     * make it a patch for pve-qemu instead of the qemu submodule
>     * use v2 of the patch with a comment by Stefan Hajnoczi
> 
>  ...uring-resubmit-when-result-is-EAGAIN.patch | 49 +++++++++++++++++++
>  debian/patches/series                         |  1 +
>  2 files changed, 50 insertions(+)
>  create mode 100644 debian/patches/extra/0002-block-io_uring-resubmit-when-result-is-EAGAIN.patch
> 
>

applied, thanks!

For others regarding bump: while this is def. correct it actually increases the
likelihood of a kernel crash with old (<< pve-kernel-5.11.22-3-pve 5.11.22-6)
kernels. As now it was either a crash or a block job error, this patch basically
converts the block job errors into crashes on bad old kernel and into avoiding
a block job error on the OK behavior of resubmitting on EAGAIN on new kernels.
Those that saw neither crash nor block job errors won't notice a difference
either way.