linux-kernel - Re: [PATCH v3 1/6] x86-64: Clean up vdso/kernel shared variables

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] [day] [month] [year] [list]

Date:	Thu, 12 May 2011 07:16:36 -0400
From:	Andrew Lutomirski <luto@....edu>
To:	Borislav Petkov <bp@...64.org>
Cc:	x86@...nel.org, linux-kernel@...r.kernel.org,
	Ingo Molnar <mingo@...e.hu>, Andi Kleen <andi@...stfloor.org>,
	Linus Torvalds <torvalds@...ux-foundation.org>,
	Nick Piggin <npiggin@...e.de>,
	"David S. Miller" <davem@...emloft.net>,
	Eric Dumazet <eric.dumazet@...il.com>,
	Peter Zijlstra <a.p.zijlstra@...llo.nl>,
	Thomas Gleixner <tglx@...utronix.de>
Subject: Re: [PATCH v3 1/6] x86-64: Clean up vdso/kernel shared variables

On Tue, May 10, 2011 at 10:48 AM, Borislav Petkov <bp@...64.org> wrote:
> On Tue, May 10, 2011 at 10:15:03AM -0400, Andy Lutomirski wrote:
>> Variables that are shared between the vdso and the kernel are
>> currently a bit of a mess.  They are each defined with their own
>> magic, they are accessed differently in the kernel, the vsyscall page,
>> and the vdso, and one of them (vsyscall_clock) doesn't even really
>> exist.
>>
>> This changes them all to use a common mechanism.  All of them are
>> delcared in vvar.h with a fixed address (validated by the linker
>> script).  In the kernel (as before), they look like ordinary
>> read-write variables.  In the vsyscall page and the vdso, they are
>> accessed through a new macro VVAR, which gives read-only access.
>>
>> The vdso is now loaded verbatim into memory without any fixups.  As a
>> side bonus, access from the vdso is faster because a level of
>> indirection is removed.
>> ---
>>  arch/x86/include/asm/vdso.h     |   14 ----------
>>  arch/x86/include/asm/vgtod.h    |    2 -
>>  arch/x86/include/asm/vsyscall.h |   12 +-------
>>  arch/x86/include/asm/vvar.h     |   52 +++++++++++++++++++++++++++++++++++++++
>>  arch/x86/kernel/time.c          |    2 +-
>>  arch/x86/kernel/tsc.c           |    4 +-
>>  arch/x86/kernel/vmlinux.lds.S   |   34 ++++++++-----------------
>>  arch/x86/kernel/vsyscall_64.c   |   46 +++++++++++++++-------------------
>>  arch/x86/vdso/Makefile          |    2 +-
>>  arch/x86/vdso/vclock_gettime.c  |    3 +-
>>  arch/x86/vdso/vdso.lds.S        |    7 -----
>>  arch/x86/vdso/vextern.h         |   16 ------------
>>  arch/x86/vdso/vgetcpu.c         |    3 +-
>>  arch/x86/vdso/vma.c             |   27 --------------------
>>  arch/x86/vdso/vvar.c            |   12 ---------
>>  15 files changed, 91 insertions(+), 145 deletions(-)
>>  create mode 100644 arch/x86/include/asm/vvar.h
>>  delete mode 100644 arch/x86/vdso/vextern.h
>>  delete mode 100644 arch/x86/vdso/vvar.c
>>

>> +/* DECLARE_VVAR(offset, type, name) */
>> +
>> +DECLARE_VVAR(0, volatile unsigned long, jiffies)
>> +DECLARE_VVAR(128, struct vsyscall_gtod_data, vsyscall_gtod_data)
>> +DECLARE_VVAR(256, int, vgetcpu_mode)
>
> Why so spread out? Maybe put each on a single cacheline instead?

They're currently on their own cachelines.  I just spread them out
because there's room and that way modification to their types is less
likely to make them conflict.

I'll add a patch at the end of the series to pack them into two
cachelines.  All of the variables are read-mostly so it should be a
win.

--Andy
--
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/