Re: [PATCH V2] usb: gadget: f_fs: don't free buffer prematurely

[John Stultz]

On Thu, Apr 25, 2019 at 9:01 AM Yang, Fei <fei.yang@xxxxxxxxx> wrote:
>
> > On Wed, Apr 24, 2019 at 9:50 AM Greg KH <gregkh@xxxxxxxxxxxxxxxxxxx> wrote:
> > >
> > > On Tue, Apr 02, 2019 at 10:26:51AM +0700, John Stultz wrote:
> > > > On Wed, Mar 20, 2019 at 11:21 AM John Stultz <john.stultz@xxxxxxxxxx> wrote:
> > > > > On Wed, Mar 20, 2019 at 9:40 AM John Stultz <john.stultz@xxxxxxxxxx> wrote:
> > > > > > On Tue, Mar 19, 2019 at 10:32 PM <fei.yang@xxxxxxxxx> wrote:
> > > > > > Elsewhere it seems the ffs driver takes effort to drop any locks
> > > > > > before calling usb_ep_dequeue(), so this seems like that should
> > > > > > be addressed, but it also seems like recent change to the dwc3
> > > > > > driver has been made to avoid sleeping in that path (see fec9095bdef4 ("usb:
> > > > > > dwc3: gadget: remove wait_end_transfer")), which may be why I'm
> > > > > > not seeing the problem with mainline (and your patch here, of coarse).
> > > > > > But that also doesn't clarify if its still a potential issue w/
> > > > > > non-dwc3 platforms.
> > > > >
> > > > > Felipe: Given Alan's point, does it make sense to mark the commits
> > > > > that remove the possible sleep from wait_event_lock_irq() in
> > > > > dwc3_gadget_ep_dequeue()  for -stable?
> > > > >
> > > > > Against 4.19.30, the following set manages to cherry-pick cleanly:
> > > > > git cherry-pick 1a22ec643580626f439c8583edafdcc73798f2fb
> > > > > git cherry-pick 09fe1f8d7e2f461275b1cdd832f2cfa5e9be346d
> > > > > git cherry-pick c3acd59014148470dc58519870fbc779785b4bf7
> > > > > git cherry-pick 7746a8dfb3f9c91b3a0b63a1d5c2664410e6498d
> > > > > git cherry-pick d5443bbf5fc8f8389cce146b1fc2987cdd229d12
> > > > > git cherry-pick d4f1afe5e896c18ae01099a85dab5e1a198bd2a8
> > > > > git cherry-pick fec9095bdef4e7c988adb603d0d4f92ee735d4a1
> > > > > # To get things building, revert modified -stable fix git revert
> > > > > 25ad17d #pick actual upstream fix replacing the previous git
> > > > > cherry-pick bd6742249b9ca918565e4e3abaa06665e587f4b5
> > > > >
> > > > > (Though I'm always a bit hesitant with -stable backports on
> > > > > subsystems I don't know well. So I'm not sure if this set is fully
> > > > > correct.)
> > > > >
> > > > > This set seems to avoid the crash on reboot I was seeing.
> > > > >
> > > > > And of course, I'm sure getting that set backported to 4.14 and
> > > > > 4.9 (and maybe even 4.4, I need to check) will be less clean.
> > > >
> > > > Also,  I just wanted to follow up on this as well. Does the above
> > > > set of cherry-picks look ok to others for 4.19-stable?  Does anyone
> > > > have suggestions on how they'd like to see backports to 4.14, 4.9 and 4.4?
> > >
> > > If they are ok, can someone send me the commits as a series of
> > > patches, as doing the above really doesn't help much :)
> >
> > Yea, so I'm happy to send that set to you for 4.19, assuming Felipe or someone else OKs it.
> > I'm still at a bit of a loss for what to do for older (4.14/4.9/4.4) -stable kernels, as its going to be difficult to backport those.
>
> The patch mentioned in the title of this thread doesn't apply to older kernels. The kernel panic was introduced in 5.0-rcX.

Right. On older kernels there was a separate issue that seems to be
dwc3 specific that was biting me in a similar way. That's what the sha
list above references.

thanks
-john