Date: Mon, 6 Jan 2020 11:32:14 -0500 From: Rich Felker <dalias@...c.org> To: d.dorau@....de Cc: musl@...ts.openwall.com Subject: Re: Issue with musl and valgrind On Mon, Jan 06, 2020 at 05:24:35PM +0100, d.dorau@....de wrote: > "Rich Felker" <dalias@...ifal.cx> schrieb am 06.01.2020 14:56:23: > [...] > > > > This is really a bug in valgrind, that it's relying on a mix of the > > dynamic linker and its own mechanism for interposing malloc. Either > > the interposition should not take place until after line 1913 of > > __dls3 in dynlink.c (this is what would happen if you did it via > > LD_PRELOAD without valgrind), or valgrind should do its own loading > > and relocation of vgpreload_memcheck-*.so independent of the tracee's > > dynamic linker. > > > > On most archs, the problem does not manifest as things are setup so > > that it's safe to call free in an unrelocated context. However > > realloc is gratuitously not safe and Adélie Linux has a patch for it > > to do it like free: > > > > > https://code.foxkit.us/adelie/packages/blob/master/user/valgrind/realloc.patch > > > > > But on MIPS, where there are no PC-relative references and everything > > goes through the GOT, this blows up. > > > > I'm not sure how practical it is to get valgrind to fix this upstream. > > In the either/or above, if the first course of action is taken, the > > mechanism should probably be by inserting a breakpoint (or valgrind's > > equivalent) at the main program's e_entry address and not doing the > > replacement hack until then. Alternatively, it could just rely on > > LD_PRELOAD working without trying to do additional tricks, but maybe > > they have a good reason they're not doing that. > > > > Rich > > I was not aware of this when debugging the crash, but after looking at > additional > debug statements and valgrind code my current understanding is that > valgrind > actually inserts > > LD_PRELOAD=" > /usr/lib/valgrind/vgpreload_core-mips32-linux.so:/usr/lib/valgrind/vgpreload_memcheck-mips32-linux.so" > > into the environment before loading the executable to test. > (coregrind/m_initimg/initimg-linux.c:123) > > 425:11:32:696 --4757:1: main Split up command line > 425:11:32:696 --4757:1: main (early_) Process Valgrind's command line > options > 425:11:32:696 --4757:1: main Create initial image > 425:11:32:712 --4757:1: initimg Loading client > 425:11:32:712 --4757:1: initimg Setup client env > 425:11:32:712 --4757:2: initimg preload_string: > 425:11:32:728 --4757:2: initimg > "/usr/lib/valgrind/vgpreload_core-mips32-linux.so:/usr/lib/valgrind/vgpreload_memcheck-mips32-linux.so" > 425:11:32:728 --4757:1: initimg Setup client stack: size will be 1048576 > > > Is my understanding correct that you would expect it to not run into > the crash I described earlier then? No, it does something explicitly wrong that causes the crash, and now I'm starting to remember. What it does is hook the mmap of vgpreload_memcheck-*.so and activate its redirections as soon as it's mapped into memory. This activation should instead be moved to e_entry point time. BTW you accidentally un-CC'd the list; I bounced the email back on to reply. Please keep replies on-list so others can follow. Rich
Powered by blists - more mailing lists
Confused about mailing lists and their use? Read about mailing lists on Wikipedia and check out these guidelines on proper formatting of your messages.