From owner-svn-src-all@FreeBSD.ORG Tue Sep 7 00:23:46 2010 Return-Path: Delivered-To: svn-src-all@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:4f8:fff6::34]) by hub.freebsd.org (Postfix) with ESMTP id 412BF10656BA; Tue, 7 Sep 2010 00:23:46 +0000 (UTC) (envelope-from rstone@FreeBSD.org) Received: from svn.freebsd.org (svn.freebsd.org [IPv6:2001:4f8:fff6::2c]) by mx1.freebsd.org (Postfix) with ESMTP id 163798FC0C; Tue, 7 Sep 2010 00:23:46 +0000 (UTC) Received: from svn.freebsd.org (localhost [127.0.0.1]) by svn.freebsd.org (8.14.3/8.14.3) with ESMTP id o870NjOh072609; Tue, 7 Sep 2010 00:23:45 GMT (envelope-from rstone@svn.freebsd.org) Received: (from rstone@localhost) by svn.freebsd.org (8.14.3/8.14.3/Submit) id o870Njtg072607; Tue, 7 Sep 2010 00:23:45 GMT (envelope-from rstone@svn.freebsd.org) Message-Id: <201009070023.o870Njtg072607@svn.freebsd.org> From: Ryan Stone Date: Tue, 7 Sep 2010 00:23:45 +0000 (UTC) To: src-committers@freebsd.org, svn-src-all@freebsd.org, svn-src-head@freebsd.org X-SVN-Group: head MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Cc: Subject: svn commit: r212281 - head/sys/vm X-BeenThere: svn-src-all@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: "SVN commit messages for the entire src tree \(except for " user" and " projects" \)" List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Tue, 07 Sep 2010 00:23:46 -0000 Author: rstone Date: Tue Sep 7 00:23:45 2010 New Revision: 212281 URL: http://svn.freebsd.org/changeset/base/212281 Log: In munmap() downgrade the vm_map_lock to a read lock before taking a read lock on the pmc-sx lock. This prevents a deadlock with pmc_log_process_mappings, which has an exclusive lock on pmc-sx and tries to get a read lock on a vm_map. Downgrading the vm_map_lock in munmap allows pmc_log_process_mappings to continue, preventing the deadlock. Without this change I could cause a deadlock on a multicore 8.1-RELEASE system by having one thread constantly mmap'ing and then munmap'ing a PROT_EXEC mapping in a loop while I repeatedly invoked and stopped pmcstat in system-wide sampling mode. Reviewed by: fabient Approved by: emaste (mentor) MFC after: 2 weeks Modified: head/sys/vm/vm_mmap.c Modified: head/sys/vm/vm_mmap.c ============================================================================== --- head/sys/vm/vm_mmap.c Mon Sep 6 23:52:04 2010 (r212280) +++ head/sys/vm/vm_mmap.c Tue Sep 7 00:23:45 2010 (r212281) @@ -579,6 +579,7 @@ munmap(td, uap) * Inform hwpmc if the address range being unmapped contains * an executable region. */ + pkm.pm_address = (uintptr_t) NULL; if (vm_map_lookup_entry(map, addr, &entry)) { for (; entry != &map->header && entry->start < addr + size; @@ -587,16 +588,23 @@ munmap(td, uap) entry->end, VM_PROT_EXECUTE) == TRUE) { pkm.pm_address = (uintptr_t) addr; pkm.pm_size = (size_t) size; - PMC_CALL_HOOK(td, PMC_FN_MUNMAP, - (void *) &pkm); break; } } } #endif - /* returns nothing but KERN_SUCCESS anyway */ vm_map_delete(map, addr, addr + size); + +#ifdef HWPMC_HOOKS + /* downgrade the lock to prevent a LOR with the pmc-sx lock */ + vm_map_lock_downgrade(map); + if (pkm.pm_address != (uintptr) NULL) + PMC_CALL_HOOK(td, PMC_FN_MUNMAP, (void *) &pkm); + vm_map_unlock_read(map); +#else vm_map_unlock(map); +#endif + /* vm_map_delete returns nothing but KERN_SUCCESS anyway */ return (0); }