Created attachment 1213447 [details] stacktrace Description of problem: During the fuzz of libdwarf I got a memory allocation failure. From the stacktrace seems that the fault is in elfutils. Version-Release number of selected component (if applicable): 0.166 How reproducible: dwarfdump $REPRODUCER Reproducer: https://github.com/asarubbo/poc/raw/master/00011-elfutils-memalloc-allocate_elf
The problem is that the ELF file claims to have 1073901568 section headers. We then try to allocate memory for the sizeof (Elf_Scn) (208) times that number. Which is 223371526144. Plus the sizeof (Elf), 264 bytes. In total 223371526408 bytes (~208GB). Which obviously fails and returns NULL. This isn't really fatal. You just get an ELF_E_NOMEM. There is already a check that scncnt <= SIZE_MAX / (sizeof (Elf_Scn) + sizeof (Elf64_Shdr)) to prevent overflow. But that still allows for 67818912035696880 headers on 64bit systems (and 15790320 on 32bit systems).
But even before that sanity check there is a call to get_shnum() which itself should sanity check that the "advertised" headers are really there. /* If the section headers were truncated, pretend none were there. */ if (ehdr.e64->e_shoff > maxsize || maxsize - ehdr.e64->e_shoff < sizeof (Elf64_Shdr) * result) result = 0; That should have triggered, but it doesn't because maxsize was set to infinity. The problem is that we only set maxsize in read_file() in case we mmap the file, but not when using pread. Posted a patch upstream: https://lists.fedorahosted.org/archives/list/elfutils-devel@lists.fedorahosted.org/message/EJWVY7TMRDEMWPAPNVU3V4MZYG5HANF2/ Does that help your case?
I'll retest. I guess my test will be done when a new release includes this fix. So if you can make a release you will make my job easier.
(In reply to Agostino Sarubbo from comment #3) > I'll retest. I guess my test will be done when a new release includes this > fix. So if you can make a release you will make my job easier. We are kind of waiting on each other :) I am afraid this isn't reason enough to do a new release, sorry. Also I would like some time for this bug and bug #1387584 to have some testing. Neither commit is even upstream yet because nobody commented, reviewed or tested them yet.
If all you need is a "release" tar ball then you could try https://fedorapeople.org/~mjw/elfutils-0.167.tar.bz2 that is elfutils 0.167 + git + the two patches for fixing bug #1387584 and bug #1388057. Any testing is appreciated.
The patch has been committed upstream and the fix will be in the next release.
elfutils-0.168-1.fc25 has been submitted as an update to Fedora 25. https://bodhi.fedoraproject.org/updates/FEDORA-2016-0cdc2d3b14
elfutils-0.168-1.fc25 has been pushed to the Fedora 25 testing repository. If problems still persist, please make note of it in this bug report. See https://fedoraproject.org/wiki/QA:Updates_Testing for instructions on how to install test updates. You can provide feedback for this update here: https://bodhi.fedoraproject.org/updates/FEDORA-2016-0cdc2d3b14
elfutils-0.168-1.fc25 has been pushed to the Fedora 25 stable repository. If problems still persist, please make note of it in this bug report.
elfutils-0.168-1.fc24 has been submitted as an update to Fedora 24. https://bodhi.fedoraproject.org/updates/FEDORA-2017-6c9a495a48
elfutils-0.168-1.fc24 has been pushed to the Fedora 24 testing repository. If problems still persist, please make note of it in this bug report. See https://fedoraproject.org/wiki/QA:Updates_Testing for instructions on how to install test updates. You can provide feedback for this update here: https://bodhi.fedoraproject.org/updates/FEDORA-2017-6c9a495a48
elfutils-0.168-1.fc24 has been pushed to the Fedora 24 stable repository. If problems still persist, please make note of it in this bug report.
Mitre assigned CVE-2016-10254 to this issue.