From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1753865Ab0IPQmE (ORCPT ); Thu, 16 Sep 2010 12:42:04 -0400 Received: from relay3.sgi.com ([192.48.152.1]:44716 "EHLO relay.sgi.com" rhost-flags-OK-OK-OK-FAIL) by vger.kernel.org with ESMTP id S1753109Ab0IPQmD (ORCPT ); Thu, 16 Sep 2010 12:42:03 -0400 To: kexec@lists.infradead.org, mingo@elte.hu Subject: [PATCH] x86: saving vmcore with non-lazy freeing of vmas Cc: linux-kernel@vger.kernel.org Message-Id: From: Cliff Wickman Date: Thu, 16 Sep 2010 11:44:02 -0500 Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org From: Cliff Wickman During the reading of /proc/vmcore the kernel is doing ioremap()/iounmap() repeatedly. And the buildup of un-flushed vm_area_struct's is causing a great deal of overhead. (rb_next() is chewing up most of that time). This solution is to provide function set_iounmap_nonlazy(). It causes a subsequent call to iounmap() to immediately purge the vma area (with try_purge_vmap_area_lazy()). With this patch we have seen the time for writing a 250MB compressed dump drop from 71 seconds to 44 seconds. Diffed against 2.6.36-rc3 Signed-off-by: Cliff Wickman --- arch/x86/include/asm/io.h | 1 + arch/x86/kernel/crash_dump_64.c | 1 + mm/vmalloc.c | 9 +++++++++ 3 files changed, 11 insertions(+) Index: linus.current/arch/x86/kernel/crash_dump_64.c =================================================================== --- linus.current.orig/arch/x86/kernel/crash_dump_64.c +++ linus.current/arch/x86/kernel/crash_dump_64.c @@ -46,6 +46,7 @@ ssize_t copy_oldmem_page(unsigned long p } else memcpy(buf, vaddr + offset, csize); + set_iounmap_nonlazy(); iounmap(vaddr); return csize; } Index: linus.current/mm/vmalloc.c =================================================================== --- linus.current.orig/mm/vmalloc.c +++ linus.current/mm/vmalloc.c @@ -517,6 +517,15 @@ static atomic_t vmap_lazy_nr = ATOMIC_IN static void purge_fragmented_blocks_allcpus(void); /* + * called before a call to iounmap() if the caller wants vm_area_struct's + * immediately freed. + */ +void set_iounmap_nonlazy(void) +{ + atomic_set(&vmap_lazy_nr, lazy_max_pages()+1); +} + +/* * Purges all lazily-freed vmap areas. * * If sync is 0 then don't purge if there is already a purge in progress. Index: linus.current/arch/x86/include/asm/io.h =================================================================== --- linus.current.orig/arch/x86/include/asm/io.h +++ linus.current/arch/x86/include/asm/io.h @@ -206,6 +206,7 @@ static inline void __iomem *ioremap(reso extern void iounmap(volatile void __iomem *addr); +extern void set_iounmap_nonlazy(void); #ifdef __KERNEL__ From mboxrd@z Thu Jan 1 00:00:00 1970 Return-path: Received: from relay3.sgi.com ([192.48.152.1] helo=relay.sgi.com) by bombadil.infradead.org with esmtp (Exim 4.72 #1 (Red Hat Linux)) id 1OwHXA-0006Uw-Oj for kexec@lists.infradead.org; Thu, 16 Sep 2010 16:42:05 +0000 Subject: [PATCH] x86: saving vmcore with non-lazy freeing of vmas Message-Id: From: Cliff Wickman Date: Thu, 16 Sep 2010 11:44:02 -0500 List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Sender: kexec-bounces@lists.infradead.org Errors-To: kexec-bounces+dwmw2=infradead.org@lists.infradead.org To: kexec@lists.infradead.org, mingo@elte.hu Cc: linux-kernel@vger.kernel.org From: Cliff Wickman During the reading of /proc/vmcore the kernel is doing ioremap()/iounmap() repeatedly. And the buildup of un-flushed vm_area_struct's is causing a great deal of overhead. (rb_next() is chewing up most of that time). This solution is to provide function set_iounmap_nonlazy(). It causes a subsequent call to iounmap() to immediately purge the vma area (with try_purge_vmap_area_lazy()). With this patch we have seen the time for writing a 250MB compressed dump drop from 71 seconds to 44 seconds. Diffed against 2.6.36-rc3 Signed-off-by: Cliff Wickman --- arch/x86/include/asm/io.h | 1 + arch/x86/kernel/crash_dump_64.c | 1 + mm/vmalloc.c | 9 +++++++++ 3 files changed, 11 insertions(+) Index: linus.current/arch/x86/kernel/crash_dump_64.c =================================================================== --- linus.current.orig/arch/x86/kernel/crash_dump_64.c +++ linus.current/arch/x86/kernel/crash_dump_64.c @@ -46,6 +46,7 @@ ssize_t copy_oldmem_page(unsigned long p } else memcpy(buf, vaddr + offset, csize); + set_iounmap_nonlazy(); iounmap(vaddr); return csize; } Index: linus.current/mm/vmalloc.c =================================================================== --- linus.current.orig/mm/vmalloc.c +++ linus.current/mm/vmalloc.c @@ -517,6 +517,15 @@ static atomic_t vmap_lazy_nr = ATOMIC_IN static void purge_fragmented_blocks_allcpus(void); /* + * called before a call to iounmap() if the caller wants vm_area_struct's + * immediately freed. + */ +void set_iounmap_nonlazy(void) +{ + atomic_set(&vmap_lazy_nr, lazy_max_pages()+1); +} + +/* * Purges all lazily-freed vmap areas. * * If sync is 0 then don't purge if there is already a purge in progress. Index: linus.current/arch/x86/include/asm/io.h =================================================================== --- linus.current.orig/arch/x86/include/asm/io.h +++ linus.current/arch/x86/include/asm/io.h @@ -206,6 +206,7 @@ static inline void __iomem *ioremap(reso extern void iounmap(volatile void __iomem *addr); +extern void set_iounmap_nonlazy(void); #ifdef __KERNEL__ _______________________________________________ kexec mailing list kexec@lists.infradead.org http://lists.infradead.org/mailman/listinfo/kexec