[Regression] snd_emu10k1 doesn't work after GCC 4.2 upgrade

[Yuriy Tsibizov]

Alexander,
I've narrowed it down to single function -- emu_vsetup (in sys/dev/sound/pci/emu10k1.c). Without -ftree-vrp it compiles without problems, with it it makes broken code. You can also compare it to the function with the same name in emu10kx.c -- in last file it compiles with or without -ftree-vrp without problems.

If you need more info, I can do more research tomorrow.

Yuriy.

P.S. Many thanks to ssb from #c on  RusNet.

-----Original Message-----
From: "Alexander Kabaev"<kabaev at gmail.com>
Sent: 26.05.07 23:48:33
To: "Yuriy Tsibizov"<Yuriy.Tsibizov at gfk.ru>
Cc: "Yuriy.Tsibizov at gfk.ru"<Yuriy.Tsibizov at gfk.ru>, "ed at fxq.nl"<ed at fxq.nl>, "current at freebsd.org"<current at freebsd.org>, "freebsd at unixfreunde.de"<freebsd at unixfreunde.de>
Subject: Re: [Regression] snd_emu10k1 doesn't work after GCC 4.2 upgrade

On Sat, 26 May 2007 23:38:58 +0400
"Yuriy Tsibizov" <Yuriy.Tsibizov at gfk.ru> wrote:

> Alexander, I'm working on it.
> The only thing I can say right now, that it's -ftree-vrp that broke
> the code -- building with -O -ftree-vrp makes broken binary, while -O
> alone does not. 
> 
> (sorry for top-posting)

Thanks. Ideally should be able to distill the case info a simple
preprocessed .i file with just enough code to demonstrate the
mis-compilation. I do not have any of the snd_emuXXX cards here myself.

-- 
Alexander Kabaev