[pve-devel] [PATCH kernel 1/5] fix #5430: cherry-pick fix for ocfs2 io_uring rw issues

[Daniel Kral]

Reported in the community forum [0].

This fixes an issue with read/write operations done on ocfs2 with
io_uring. This has caused QEMU guests to be unable to determine the file
format at [1] because of an unsuccessful read and therefore could not
boot, which has been resolved with this patch.

This patch is already merged in Jens Axboe's linux-block tree and also
merged in the mainline v6.12 prepatch kernels:

> # git tag --contains c0a9d496e0fece67db777bd48550376cf2960c47
> v6.12-rc1
> v6.12-rc2
> v6.13-rc3

[0] https://forum.proxmox.com/threads/140273/post-702007
[1] https://elixir.bootlin.com/qemu/v9.0.2/source/block.c#L1031

Signed-off-by: Daniel Kral <d.kral at proxmox.com>
---
 ...t-EOPNOTSUPP-for-IOCB_NOWAIT-like-EA.patch | 45 +++++++++++++++++++
 1 file changed, 45 insertions(+)
 create mode 100644 patches/kernel/0021-io_uring-rw-treat-EOPNOTSUPP-for-IOCB_NOWAIT-like-EA.patch

diff --git a/patches/kernel/0021-io_uring-rw-treat-EOPNOTSUPP-for-IOCB_NOWAIT-like-EA.patch b/patches/kernel/0021-io_uring-rw-treat-EOPNOTSUPP-for-IOCB_NOWAIT-like-EA.patch
new file mode 100644
index 0000000..7187c94
--- /dev/null
+++ b/patches/kernel/0021-io_uring-rw-treat-EOPNOTSUPP-for-IOCB_NOWAIT-like-EA.patch
@@ -0,0 +1,45 @@
+From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
+From: Jens Axboe <axboe at kernel.dk>
+Date: Tue, 10 Sep 2024 08:30:57 -0600
+Subject: [PATCH] io_uring/rw: treat -EOPNOTSUPP for IOCB_NOWAIT like -EAGAIN
+
+Some file systems, ocfs2 in this case, will return -EOPNOTSUPP for
+an IOCB_NOWAIT read/write attempt. While this can be argued to be
+correct, the usual return value for something that requires blocking
+issue is -EAGAIN.
+
+A refactoring io_uring commit dropped calling kiocb_done() for
+negative return values, which is otherwise where we already do that
+transformation. To ensure we catch it in both spots, check it in
+__io_read() itself as well.
+
+Reported-by: Robert Sander <r.sander at heinlein-support.de>
+Link: https://fosstodon.org/@gurubert@mastodon.gurubert.de/113112431889638440
+Cc: stable at vger.kernel.org
+Fixes: a08d195b586a ("io_uring/rw: split io_read() into a helper")
+Signed-off-by: Jens Axboe <axboe at kernel.dk>
+(cherry picked from commit c0a9d496e0fece67db777bd48550376cf2960c47)
+Signed-off-by: Daniel Kral <d.kral at proxmox.com>
+---
+ io_uring/rw.c | 8 ++++++++
+ 1 file changed, 8 insertions(+)
+
+diff --git a/io_uring/rw.c b/io_uring/rw.c
+index c3c154790e45..ed7f67097572 100644
+--- a/io_uring/rw.c
++++ b/io_uring/rw.c
+@@ -825,6 +825,14 @@ static int __io_read(struct io_kiocb *req, unsigned int issue_flags)
+ 
+ 	ret = io_iter_do_read(rw, &s->iter);
+ 
++	/*
++	 * Some file systems like to return -EOPNOTSUPP for an IOCB_NOWAIT
++	 * issue, even though they should be returning -EAGAIN. To be safe,
++	 * retry from blocking context for either.
++	 */
++	if (ret == -EOPNOTSUPP && force_nonblock)
++		ret = -EAGAIN;
++
+ 	if (ret == -EAGAIN || (req->flags & REQ_F_REISSUE)) {
+ 		req->flags &= ~REQ_F_REISSUE;
+ 		/*
-- 
2.39.5