Re: [PATCH v5.10.y 1/1] Revert "io_uring: reinforce cancel on flush during exit"

[Lee Jones]

On Tue, 02 Nov 2021, Greg KH wrote:

> On Tue, Nov 02, 2021 at 03:49:30PM +0000, Lee Jones wrote:
> > This reverts commit 88dbd085a51ec78c83dde79ad63bca8aa4272a9d.
> > 
> > Causes the following Syzkaller reported issue:
> > 
> > LINK: https://syzkaller.appspot.com/bug?extid=b0003676644cf0d6acc4
> 
> "Link:"?

Sure.

> > Reported-by: syzbot+b0003676644cf0d6acc4@xxxxxxxxxxxxxxxxxxxxxxxxx
> > Signed-off-by: Lee Jones <lee.jones@xxxxxxxxxx>
> 
> Why does the patch here cause this error?  Is it due to the backport
> being different than what went into Linus's tree, or something else?

The patch is the same.

I need to do some more testing, but I assume it *was* also broken in
Mainline until the whole section was rewritten.  Unfortunately, the
patch which does that does not apply to v5.10.y.  Again, I can look
into this, but I'm not quite sure how far the rabbit hole goes.

> The original commit did fix a real issue, what should we do now to
> resolve that issue in 5.10.y instead?

Ideally I'd like the original author and/or the domain experts to have
a look and chime in here.

-- 
Lee Jones [李琼斯]
Senior Technical Lead - Developer Services
Linaro.org │ Open source software for Arm SoCs
Follow Linaro: Facebook | Twitter | Blog