DMARC-Filter: OpenDMARC Filter v1.3.2 mx1.redhat.com DC6542EF18A
Date: Tue, 28 Mar 2017 16:54:13 +0200
From: Oleg Nesterov <oleg@redhat.com>
To: Andrew Morton <akpm@linux-foundation.org>,
        Andy Lutomirski <luto@kernel.org>,
        Linus Torvalds <torvalds@linux-foundation.org>
Cc: Denys Vlasenko <dvlasenk@redhat.com>, "H. Peter Anvin" <hpa@zytor.com>,
        Ingo Molnar <mingo@redhat.com>,
        Jan Kratochvil <jan.kratochvil@redhat.com>,
        Pedro Alves <palves@redhat.com>, Thomas Gleixner <tglx@linutronix.de>,
        x86@kernel.org, linux-kernel@vger.kernel.org
Subject: [PATCH 0/1] get_nr_restart_syscall() should return
 __NR_ia32_restart_syscall if __USER32_CS
Message-ID: <20170328145413.GA3164@redhat.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
User-Agent: Mutt/1.5.24 (2015-08-30)
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 860
Lines: 30

Hello,

get_nr_restart_syscall() is still buggy, TS_I386_REGS_POKED can't
really help and should probably die.

The fix just adds the __USER32_CS check, but perhaps we can avoid
these "fundamentally broken" checks altogether?

Is __NR_ia32_restart_syscall/__NR_restart_syscall the part of ABI?
OK, we probaly can't remove them, at least right now. But what if
we simply add the new syscall number,

	#define __NR_new_restart_syscall	383
	#define __NR_ia32_new_restart_syscall	383

so that it doesn't depends on bitness and we can just do

	static inline unsigned long get_nr_restart_syscall(const struct pt_regs *regs)
	{
		BUILD_BUG_ON(__NR_ia32_new_restart_syscall != __NR_new_restart_syscall);
	#ifdef CONFIG_X86_X32_ABI
		return __NR_new_restart_syscall | (regs->orig_ax & __X32_SYSCALL_BIT);
	#else
		return __NR_new_restart_syscall;
	#endif
	}

?

Oleg.