Date: Tue, 7 Aug 2007 15:44:21 +0200 From: Tijl Coosemans <tijl@ulyssis.org> To: freebsd-current@freebsd.org Cc: Kostik Belousov <kostikbel@gmail.com>, wine-freebsd@hub.org Subject: Re: mmap(2) MAP_FIXED isn't thread-safe (+testcase) Message-ID: <200708071544.23556.tijl@ulyssis.org> In-Reply-To: <20070807103910.GI2738@deviant.kiev.zoral.com.ua> References: <200708051656.50168.tijl@ulyssis.org> <200708062032.10514.tijl@ulyssis.org> <20070807103910.GI2738@deviant.kiev.zoral.com.ua>
next in thread | previous in thread | raw e-mail | index | archive | help
On Tuesday 07 August 2007 12:39:11 Kostik Belousov wrote: > On Mon, Aug 06, 2007 at 08:32:08PM +0200, Tijl Coosemans wrote: >> I don't think you can do this. Now, when vm_map_find() is called >> with find_space=FALSE, it will delete any existing mapping instead >> of failing if the address is not available. Have you checked all >> locations where this function is called to make sure this isn't >> harmful? > > Yes, I looked over this when writing the patch. I think this is what > actually supposed to happen in that case. > > Anyway, to not diverge from the old behaviour when fixing the issue, > please test patch below. I do not like introducing recursive locks > where it is quite easy to avoid them. Works for me. I've added it to the Wine wiki to give it wider testing.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200708071544.23556.tijl>