linux-kernel - Re: [GIT PULL, v4] S+core architecture (arch/score) support

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

Message-Id: <200908310956.33802.arnd@arndb.de>
Date:	Mon, 31 Aug 2009 09:56:33 +0000
From:	Arnd Bergmann <arnd@...db.de>
To:	liqin.chen@...plusct.com
Cc:	linux-arch@...r.kernel.org, linux-kernel@...r.kernel.org,
	torvalds@...ux-foundation.org
Subject: Re: [GIT PULL, v4] S+core architecture (arch/score) support

On Monday 31 August 2009 07:48:33 liqin.chen@...plusct.com wrote:
> Arnd Bergmann <arnd@...db.de> 写于 2009-08-27 22:00:37:
> 
> > 
> > I understand your problem here, and I think it would be good
> > to have second opinion from someone else here. The four
> > #defines you mention are roughly in reverse order of importance.
> > 
> > I think you really need to remove __ARCH_WANT_SYSCALL_DEPRECATED
> > and get that working. Please tell us if some system call is
> > giving you problems and we can find a solution together.
> > 
> > Removing __ARCH_WANT_SYSCALL_OFF_T should not be too hard, because
> > glibc uses the loff_t versions by default anyway. I have not tried
> > it though, so it may be more complicated than I thought.
> > 
> 
> in __ARCH_WANT_SYSCALL_DEPRECATED,
> asm-generic/unistd.h haven't define __NR_ipc and __NR__llseek
> which used in glibc.

__NR_ipc definitely should not be defined, it is a multiplexer
for other system calls that have their own number, e.g. msgctl
and semop. Look at e.g. how parisc does it in glibc.

I'm not sure what the problem with __NR_llseek could be, AFAICT
it is defined in asm-generic/unistd.h because you take the
default __BITSPERLONG=32 definition. You should however not
see __NR_lseek, which is used only on 64 bit systems these days.

> Now we have maintained 3 sets of syscall API code use for glibc.
> 1: Old syscall interface, which linux/score system work well.

Yes, and I guess this will continue to work, but it will never
work with upstream kernels.

> 2: New syscall interface, define all 4 macro __ARCH_WANT_SYSCALL_XXX,
>         linux/score system work now, but we still debug the glibc code.
> 3: New syscall interface, without define __ARCH_WANT_SYSCALL_DEPRECATED
>         and __ARCH_WANT_SYSCALL_OFF_T, linux/score system couldn't work.
>         because glibc have many place need to fix.

Do you have a patch against the glibc git tree for score so I can look
at what needs to be done? None of the changes should be specific to
score after all, ideally it should all work without any glibc changes,
though I realize that we are not there yet.

	Arnd <><
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/