Re: 4.5.1: UBSAN: Undefined behaviour in ./arch/x86/include/asm/atomic.h:156:9

[Greg KH <greg-U8xfFu+wG4EAvxtiuMwx3w@public.gmane.org>]

On Thu, Apr 21, 2016 at 01:17:19PM +0200, Martin MOKREJŠ wrote:
> Greg KH wrote:
> > On Thu, Apr 21, 2016 at 11:18:16AM +0200, Martin MOKREJŠ wrote:
> > > Hi Greg,
> > > 
> > > thank you for your answer.
> > > 
> > > Greg KH wrote:
> > > > On Thu, Apr 21, 2016 at 12:22:49AM +0200, Martin MOKREJŠ wrote:
> > > > > Hi,
> > > > >   I am not certain to to forward this to, so I am trying linux-usb and ext4. Please add relevant people/lists, thank you.
> > > > > 
> > > > > # dmesg | grep "UBSAN: Undefined behaviour"
> > > > > [    2.638843] UBSAN: Undefined behaviour in drivers/usb/host/ehci-hub.c:873:47
> > > > > [    8.553620] UBSAN: Undefined behaviour in fs/ext4/mballoc.c:2612:15
> > > > > [   25.341048] UBSAN: Undefined behaviour in fs/ext4/mballoc.c:1274:11
> > > > > [   41.460542] UBSAN: Undefined behaviour in ./arch/x86/include/asm/atomic.h:156:9
> > > > > #
> > > > > 
> > > > > Full dmesg is attached.
> > > > > 
> > > > > Hope this helps,
> > > > > Martin
> > > > 
> > > > > [    0.000000] Linux version 4.5.1-default-pciehp (root@vostro) (gcc version 5.3.0 (Gentoo 5.3.0 p1.0, pie-0.6.5) ) #1 SMP Thu Apr 14 14:17:48 CEST 2016
> > > > 
> > > > Please try 4.6-rc4, a number of UBSAN issues have been fixed there.
> > > 
> > > They are still present:
> > > 
> > > # dmesg | grep "UBSAN: Undefined behaviour"
> > > [    3.113405] UBSAN: Undefined behaviour in drivers/usb/host/ehci-hub.c:877:47
> > > [    5.678949] UBSAN: Undefined behaviour in fs/ext4/mballoc.c:2619:15
> > > [   18.345599] UBSAN: Undefined behaviour in fs/ext4/mballoc.c:1281:11
> > > [   34.544363] UBSAN: Undefined behaviour in ./arch/x86/include/asm/atomic.h:156:9
> > 
> > Ah, sorry about that, turns out this is a bug in the tool, not in the
> > code, see the thread from a few weeks ago in the linux-usb archives with
> > the subject:
> > 	Subject: function ehci_hub_control in ehci-hub.c
> > for all of the details.
> 
> Thank you. I am not a programmer so I only infer from this thread I should disable
> the UBSAN in .config. The kernel code is technically correct but triggers false alarm.
> Maybe the UBSAN could have a whitelist of falsely matching locations?
> 
> https://www.mail-archive.com/linux-usb-u79uwXL29TY76Z2rM5mHXA@public.gmane.org/msg72438.html

It probably should be fixed to not report false things like this.

> Did anybody check the ext4 code or is it also just believed the UBSAN complaint is
> just broken and code changes are not necessary?

I'm not an ext4 developer, I'll leave that to them  :)

thanks,

greg k-h
--
To unsubscribe from this list: send the line "unsubscribe linux-usb" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html