linux-kernel - Re: linux-next: build failure after merge of the akpm tree

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 for Android: free password hash cracker in your pocket

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

Message-ID: <s5hmvs0uytw.wl-tiwai@suse.de>
Date:	Wed, 20 Jan 2016 15:09:47 +0100
From:	Takashi Iwai <tiwai@...e.de>
To:	Vinod Koul <vinod.koul@...ux.intel.com>
Cc:	Andrew Morton <akpm@...ux-foundation.org>,
	Stephen Rothwell <sfr@...b.auug.org.au>,
	linux-next@...r.kernel.org, linux-kernel@...r.kernel.org,
	Pierre-Louis Bossart <pierre-louis.bossart@...ux.intel.com>
Subject: Re: linux-next: build failure after merge of the akpm tree

On Sat, 16 Jan 2016 09:51:29 +0100,
Takashi Iwai wrote:
> 
> On Sat, 16 Jan 2016 00:14:01 +0100,
> Andrew Morton wrote:
> > 
> > On Sat, 16 Jan 2016 10:00:47 +1100 Stephen Rothwell <sfr@...b.auug.org.au> wrote:
> > 
> > > Hi Andrew,
> > > > +#define SNDRV_COMPRESS_GET_CODEC_CAPS	_IOWR('C', 0x11, struct snd_compr_caps)
> > > >  #define SNDRV_COMPRESS_SET_PARAMS	_IOW('C', 0x12, struct snd_compr_params)
> > > >  #define SNDRV_COMPRESS_GET_PARAMS	_IOR('C', 0x13, struct snd_codec)
> > > >  #define SNDRV_COMPRESS_SET_METADATA	_IOW('C', 0x14,\
> > > > 
> > > > then it compiles OK.  So there's something special about `struct
> > > > snd_compr_codec_caps' which is confusing the compiler.
> > > 
> > > Its very big (hint, I think it is bigger than 1 << 13).
> > 
> > ah-hah.
> > 
> > sizeof(struct snd_compr_codec_caps) is 11144 on x86_64
> > allyesconfig, but x86_64 uses (1<<14), not (1<<13) - x86_64 has more
> > space in the ioctl payload.
> > 
> > 
> > So.  Sound guys: your struct snd_compr_codec_caps is too large for
> > powerpc ioctls.
> 
> Ouch, this was what I was concerned.  I thought we did calculate it at
> the time writing API, alas, we failed.  We need to do math drills
> again...
> 
> There are a few ways to fix this, but all are not comfortable.
> 
> A. Disable compress API for powerpc.
> 
> B. Disable this ioctl for powerpc and keep it for others while
>    providing yet a new ioctl for the similar functionality.
> 
>   B1. Pass the same struct via a pointer: this needs a special
>       compat_ioctl handling
>   B2. Reduce the size of struct snd_compr_codec_caps, e.g. set
>       MAX_NUM_CODEC_DESCRIPTORS to 16.
> 
> C. What else?

Vinod, what's your take?


Takashi