[Kernel] Nuvoton NPCM7xx unbind FIU issue

[George Hung (洪忠敬)]

> -----Original Message-----
> From: Joel Stanley <joel at jms.id.au>
> Sent: Wednesday, April 28, 2021 4:23 PM
> To: George Hung (洪忠敬) <George.Hung at quantatw.com>
> Cc: OpenBMC Maillist <openbmc at lists.ozlabs.org>; William Kennington
> <wak at google.com>; Brandon Kim <brandonkim at google.com>; Benjamin
> Fair <benjaminfair at google.com>; Fran Hsu (徐誌謙)
> <Fran.Hsu at quantatw.com>
> Subject: Re: [Kernel] Nuvoton NPCM7xx unbind FIU issue
> 
> On Wed, 28 Apr 2021 at 01:15, George Hung (洪忠敬)
> <George.Hung at quantatw.com> wrote:
> >
> > Hi Joel,
> >
> > For kernel v5.x, we found that when we unbind FIU module at the first
> time, it would cause kernel warning like that:
> >
> > ------------[ cut here ]------------
> > WARNING: CPU: 0 PID: 2174 at lib/refcount.c:190
> > refcount_sub_and_test_checked+0x60/0xbc
> > refcount_t: underflow; use-after-free.
> > Modules linked in:
> > CPU: 0 PID: 2174 Comm: gbs-sysinit.sh Not tainted
> > 5.4.80-ebad8cd-dirty-c46444d #1 Hardware name: NPCM7XX Chip family
> 
> Hopefully you're not using 5.4.80 anywhere :)
> 
> >
> > And there's a fix on kernel github for now:
> > https://apc01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgith
> >
> ub.com%2Ftorvalds%2Flinux%2Fcommit%2F794aaf01444d4e765e2b067cba0
> 1cc69c
> >
> 1c68ed9&data=04%7C01%7CGeorge.Hung%40quantatw.com%7C27233e
> 3f0d6046
> >
> e54c7b08d90a1eed0d%7C179b032707fc4973ac738de7313561b2%7C1%7C0%
> 7C637551
> >
> 950217698606%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJ
> QIjoiV2luM
> >
> zIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C1000&sdata=UbLX%2BUAT0sd
> FD5gn4
> > ZtwyKV7WlsykEngIZpcrHWClE4%3D&reserved=0
> >
> > Could you help to pull this fix to current OpenBMC kernel repo. to fix this
> issue ?
> 
> Sure. I've applied this change to dev-5.10.

I found this commit, thanks a lot.

Best Regards 
George Hung

> 
> I imagine it will show up in the stable tree shortly too, so if you are really
> using v5.4 I recommend merging in the latest stable tree when that happens.
> 
> Cheers,
> 
> Joel