linux-kernel - Re: [syzbot] general protection fault in gadget

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <20210416174604.GA47856@rowland.harvard.edu>
Date:   Fri, 16 Apr 2021 13:46:04 -0400
From:   Alan Stern <stern@...land.harvard.edu>
To:     Anirudh Rayabharam <mail@...rudhrb.com>
Cc:     Dmitry Vyukov <dvyukov@...gle.com>,
        syzbot <syzbot+eb4674092e6cc8d9e0bd@...kaller.appspotmail.com>,
        Andrey Konovalov <andreyknvl@...il.com>,
        Felipe Balbi <balbi@...nel.org>,
        Dan Carpenter <dan.carpenter@...cle.com>,
        Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
        LKML <linux-kernel@...r.kernel.org>,
        USB list <linux-usb@...r.kernel.org>,
        syzkaller-bugs <syzkaller-bugs@...glegroups.com>
Subject: Re: [syzbot] general protection fault in gadget_setup

On Fri, Apr 16, 2021 at 10:35:20PM +0530, Anirudh Rayabharam wrote:
> On Fri, Apr 16, 2021 at 11:27:34AM -0400, Alan Stern wrote:
> > Actually, I wanted to move this emulation code into a new subroutine and 
> > then call that subroutine from _both_ places.  Would you like to write 
> 
> Does it really need to be called from both places?

You know, I was going to say Yes, but now I think you're right; it's not 
needed in dummy_udc_stop.  This is because core.c always calls 
usb_gadget_disconnect before usb_gadget_udc_stop.  And we can rely on 
this behavior; it's obviously necessary to disconnect from the host 
before stopping the UDC driver.

On the other hand, while checking that fact I noticed that 
soft_connect_store in core.c doesn't call synchronize_irq in between the 
other two, the way usb_gadget_remove_driver does.  That seems like a bug 
-- if it's necessary to synchronize with the IRQ handler on one path, it 
should be necessary on the other path as well.  But that's a matter for 
a separate patch.

Alan Stern

> > and submit a patch that does this?
> 
> Sure! I will do that.
> 
> Thanks!
> 
> 	- Anirudh.