From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1754967Ab0CLHPn (ORCPT ); Fri, 12 Mar 2010 02:15:43 -0500 Received: from fgwmail6.fujitsu.co.jp ([192.51.44.36]:48801 "EHLO fgwmail6.fujitsu.co.jp" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753989Ab0CLHPl (ORCPT ); Fri, 12 Mar 2010 02:15:41 -0500 X-SecurityPolicyCheck-FJ: OK by FujitsuOutboundMailChecker v1.3.1 From: KOSAKI Motohiro To: Sharyathi Nagesh , Ulrich Drepper , Thomas Gleixner , Peter Zijlstra , Ingo Molnar , LKML Subject: [PATCH] sched: sched_getaffinity() allow less than NR_CPUS length Cc: kosaki.motohiro@jp.fujitsu.com Message-Id: <20100312161316.9520.A69D9226@jp.fujitsu.com> MIME-Version: 1.0 Content-Type: text/plain; charset="US-ASCII" Content-Transfer-Encoding: 7bit X-Mailer: Becky! ver. 2.50.07 [ja] Date: Fri, 12 Mar 2010 16:15:36 +0900 (JST) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Recently, some distro decided to use NR_CPUS=4096 for mysterious reason. Unfortunately, glibc sched interface have following definition. # define __CPU_SETSIZE 1024 # define __NCPUBITS (8 * sizeof (__cpu_mask)) typedef unsigned long int __cpu_mask; typedef struct { __cpu_mask __bits[__CPU_SETSIZE / __NCPUBITS]; } cpu_set_t; It mean, if NR_CPUS is bigger than 1024, cpu_set_t can makes ABI issue. More recently, Sharyathi Nagesh reported following test program makes misterious syscall failure. ----------------------------------------------------------------------- #define _GNU_SOURCE #include #include #include int main() { cpu_set_t set; if (sched_getaffinity(0, sizeof(cpu_set_t), &set) < 0) printf("\n Call is failing with:%d", errno); } ----------------------------------------------------------------------- Because the kernel assue len argument of sched_getaffinity() is bigger than NR_CPUS. but now it is not correct. Now we faced annoying dilemma. (1) if we change glibc's __CPU_SETSIZE definition, we lost binary compatibility of _all_ application. (2) if we don't change it, we also lost binary compatibility of Sharyathi's use case. Then, I would propse to change the rule of the len argument of sched_getaffinity(). Old: len should be bigger than NR_CPUS New: len should be bigger than maximum possible cpu id It makes following behavior (A) In the real 4096 cpus machine, the above test program still return -EINVAL. (B) NR_CPUS=4096 but the machine have less than 1024 cpus (almost machines in the world), the above can run successfully. Fortunatelly, BIG SGI machine is mainly used for HPC use case. It mean they can rewrite their program by themself. IOW we can think they don't annoy this issue. Reported-by: Sharyathi Nagesh Signed-off-by: KOSAKI Motohiro Cc: Ulrich Drepper Cc: Thomas Gleixner Cc: Peter Zijlstra Cc: Ingo Molnar --- kernel/sched.c | 10 +++++++--- 1 files changed, 7 insertions(+), 3 deletions(-) diff --git a/kernel/sched.c b/kernel/sched.c index 150b698..34217f1 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -4902,7 +4902,9 @@ SYSCALL_DEFINE3(sched_getaffinity, pid_t, pid, unsigned int, len, int ret; cpumask_var_t mask; - if (len < cpumask_size()) + if (len < nr_cpu_ids) + return -EINVAL; + if (len & (sizeof(unsigned long)-1)) return -EINVAL; if (!alloc_cpumask_var(&mask, GFP_KERNEL)) @@ -4910,10 +4912,12 @@ SYSCALL_DEFINE3(sched_getaffinity, pid_t, pid, unsigned int, len, ret = sched_getaffinity(pid, mask); if (ret == 0) { - if (copy_to_user(user_mask_ptr, mask, cpumask_size())) + int retlen = min(len, cpumask_size()); + + if (copy_to_user(user_mask_ptr, mask, retlen)) ret = -EFAULT; else - ret = cpumask_size(); + ret = retlen; } free_cpumask_var(mask); -- 1.6.5.2