qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
From: Christian Schoenebeck <qemu_oss@crudebyte.com>
To: qemu-devel@nongnu.org
Cc: Greg Kurz <groug@kaod.org>
Subject: Re: [PATCH 0/4] 9pfs: Fix ftruncate-after-unlink
Date: Tue, 11 Mar 2025 11:51:50 +0100	[thread overview]
Message-ID: <3436766.7mbngL4j9h@silver> (raw)
In-Reply-To: <20250310171101.138380-1-groug@kaod.org>

On Monday, March 10, 2025 6:10:57 PM CET Greg Kurz wrote:
> QEMU 9.2 already fixed the long standing limitation of failing fstat() on
> unlinked files. This series does something similar for ftruncate().
> 
> The following program can be straced inside the guest with a shared fs in
> passthrough mode over 9p2000.L.
> 
> int main(void)
> {
> 	struct stat st;
> 	int fd = creat("./foo", 0000);
> 
> 	ftruncate(fd, 100);
> 	unlink("./foo");
> 	ftruncate(fd, 1000);
> }
> 
> Before :
> 
> creat("./foo", 000)                     = 3
> ftruncate(3, 100)                       = -1 EACCES (Permission denied)
> unlink("./foo")                         = 0
> ftruncate(3, 1000)                      = -1 ENOENT (No such file or directory)
> 
> After :
> 
> creat("./foo", 000)                     = 3
> ftruncate(3, 100)                       = 0
> unlink("./foo")                         = 0
> ftruncate(3, 1000)                      = 0
> 
> Christian,
> 
> I'm not familiar enough with the latest changes to write a proper test
> for this case and I don't have enough cycles to learn. I'm sorry for that
> but I guess it will be a lot easier for you and I'll review.

With the following test client patch applied:

  https://lore.kernel.org/all/E1trx7R-002JEJ-0l@kylie.crudebyte.com/

then something like this should do it:

diff --git a/tests/qtest/virtio-9p-test.c b/tests/qtest/virtio-9p-test.c
index f515a9bb15..d15721e4b2 100644
--- a/tests/qtest/virtio-9p-test.c
+++ b/tests/qtest/virtio-9p-test.c
@@ -736,6 +736,12 @@ static void fs_use_after_unlink(void *obj, void *data,
         .data = buf
     }).count;
     g_assert_cmpint(count, ==, write_count);
+    tsetattr({ /* truncate file to (arbitrarily chosen) size 2001 */
+        .client = v9p, .fid = fid_file, .attr = (v9fs_attr) {
+            .valid = P9_SETATTR_SIZE,
+            .size = 2001
+        }
+    });
 }
 
 static void cleanup_9p_local_driver(void *data)

/Christian

> Cheers,
> 
> --
> Greg
> 
> Greg Kurz (4):
>   9pfs: local : Introduce local_fid_fd() helper
>   9pfs: Don't use file descriptors in core code
>   9pfs: Introduce ftruncate file op
>   9pfs: Introduce futimens file op
> 
>  fsdev/file-op-9p.h |  5 +++++
>  hw/9pfs/9p-local.c | 49 ++++++++++++++++++++++++++++++++++------------
>  hw/9pfs/9p-synth.c | 22 +++++++++++++++++++++
>  hw/9pfs/9p-util.h  |  1 +
>  hw/9pfs/9p.c       | 21 +++++++++++++++-----
>  hw/9pfs/cofs.c     | 37 ++++++++++++++++++++++++++++++++++
>  hw/9pfs/coth.h     |  4 ++++
>  7 files changed, 122 insertions(+), 17 deletions(-)
> 
> 




      parent reply	other threads:[~2025-03-11 10:52 UTC|newest]

Thread overview: 14+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2025-03-10 17:10 [PATCH 0/4] 9pfs: Fix ftruncate-after-unlink Greg Kurz
2025-03-10 17:10 ` [PATCH 1/4] 9pfs: local : Introduce local_fid_fd() helper Greg Kurz
2025-03-11 10:58   ` Christian Schoenebeck
2025-03-11 14:01     ` Greg Kurz
2025-03-10 17:10 ` [PATCH 2/4] 9pfs: Don't use file descriptors in core code Greg Kurz
2025-03-11 11:13   ` Christian Schoenebeck
2025-03-11 14:03     ` Greg Kurz
2025-03-11 14:13       ` Greg Kurz
2025-03-11 14:18       ` Christian Schoenebeck
2025-03-10 17:11 ` [PATCH 3/4] 9pfs: Introduce ftruncate file op Greg Kurz
2025-03-11 11:19   ` Christian Schoenebeck
2025-03-10 17:11 ` [PATCH 4/4] 9pfs: Introduce futimens " Greg Kurz
2025-03-11 11:25   ` Christian Schoenebeck
2025-03-11 10:51 ` Christian Schoenebeck [this message]

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=3436766.7mbngL4j9h@silver \
    --to=qemu_oss@crudebyte.com \
    --cc=groug@kaod.org \
    --cc=qemu-devel@nongnu.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).