Hi Miklos,
After applying your attached patch on top of linux-next next-20240924
tag, issue cannot be reproduced running the same repro binary.
Regards,
Yi Lai
On Tue, Sep 24, 2024 at 10:58:31AM +0200, Miklos Szeredi wrote:
> On Tue, 24 Sept 2024 at 01:48, Joanne Koong <joannelkoong@xxxxxxxxx> wrote:
> > So maybe just clear_bit(FR_PENDING, &req->flags) before we call
> > fuse_request_end() is the best.
>
> Agreed. Attached patch should fix it.
>
> Yi, can you please verify?
>
> Thanks,
> Miklos
> From fcd2d9e1fdcd7cada612f2e8737fb13a2bce7d0e Mon Sep 17 00:00:00 2001
> From: Miklos Szeredi <mszeredi@xxxxxxxxxx>
> Date: Tue, 24 Sep 2024 10:47:23 +0200
> Subject: fuse: clear FR_PENDING if abort is detected when sending request
>
> The (!fiq->connected) check was moved into the queuing method resulting in
> the following:
>
> Fixes: 5de8acb41c86 ("fuse: cleanup request queuing towards virtiofs")
> Reported-by: Lai, Yi <yi1.lai@xxxxxxxxxxxxxxx>
> Closes: https://lore.kernel.org/all/ZvFEAM6JfrBKsOU0@ly-workstation/
> Signed-off-by: Miklos Szeredi <mszeredi@xxxxxxxxxx>
> ---
> fs/fuse/dev.c | 1 +
> 1 file changed, 1 insertion(+)
>
> diff --git a/fs/fuse/dev.c b/fs/fuse/dev.c
> index cdf925e0b317..53c4569d85a4 100644
> --- a/fs/fuse/dev.c
> +++ b/fs/fuse/dev.c
> @@ -295,6 +295,7 @@ static void fuse_dev_queue_req(struct fuse_iqueue *fiq, struct fuse_req *req)
> } else {
> spin_unlock(&fiq->lock);
> req->out.h.error = -ENOTCONN;
> + clear_bit(FR_PENDING, &req->flags);
> fuse_request_end(req);
> }
> }
> --
> 2.46.0
>
