Issue metadata
Sign in to add a comment
|
Security: cherry-pick PDFium tiff security fixes to the Chrome OS tiff repo. |
||||||||||||||||||||||
Issue descriptionAccording to Lei, the fixes are: " The security bugs are 0006 for bug 618267 and 0017 for bug 681300 . " those patches are from https://pdfium-review.googlesource.com/c/3117/
,
Mar 24 2017
,
Mar 25 2017
,
Mar 25 2017
,
Mar 27 2017
,
Mar 29 2017
jorgelo@, sorry to bother you again. I wonder if you know who should own this.
,
Mar 29 2017
What is the main tiff repo? npm@ has been sending patches upstream to what he thought was the official repo.
,
Mar 29 2017
Sorry, by "main tiff repo" here I meant the main Chrome OS tiff package at https://cs.corp.google.com/chromeos_public/src/third_party/portage-stable/media-libs/tiff/tiff-4.0.7.ebuild.
,
Mar 29 2017
Unless you patch the package, you have to wait for it to be fixed upstream, no? PDF plugin is not the main component here, adding the one I think matches the description.
,
Mar 29 2017
This bug tracks applying the non-upstreamed tiff patches that PDFium developed, on the Chrome OS copy of tiff. Portage, the Chrome OS build system, has an easy way to apply patches to pacakges, even when those patches are not upstream. If you want to file another bug to upstream those packages, that's fine, but that's not work tracked on this bug.
,
Mar 29 2017
Oh ok, didn't see any patches on the link in #8, so that's why I asked. The CLs for 0006, 0017 are: https://codereview.chromium.org/2284063002 https://pdfium-review.googlesource.com/c/2355/
,
Apr 3 2017
,
Apr 14 2017
bugbot didn't update this, but it's been fixed here: https://chromium-review.googlesource.com/475630
,
Apr 14 2017
and i don't plan on backporting to M58 due to higher risk here ... these patches aren't in upstream, and they've only been tested in the pdfium build env (which means being built with clang and newer C++ standards). Gentoo already saw some breakage that i had to fix up in the patches.
,
Apr 15 2017
,
Apr 18 2017
Agree with not merging upstream. But could you explain what breakage you got in Gentoo? If it is just the .patch files not applying directly, that's probably because we have additional changes... Otherwise let us know if there is something we should fix in the code.
,
Apr 18 2017
0006-HeapBufferOverflow-ChopUpSingleUncompressedStrip.patch is pretty bad: it declares an internal func in the public header (tiffio.h) instead of the private one (tiffiop.h), and it's implemented in C++ instead of C (which all of libtiff is written in), and it puts the func definition in a source file that isn't part of libtiff which means it'd never work on vanilla libtiff. 0013-validate-refblackwhite.patch requires newer C standard by placing "int i" into the for loop, and it doesn't include math.h for the isnan func. really should have the patches sent to upstream libtiff.
,
Jul 22 2017
This bug has been closed for more than 14 weeks. Removing security view restrictions. For more details visit https://www.chromium.org/issue-tracking/autotriage - Your friendly Sheriffbot
,
Jan 22 2018
,
Jun 21 2018
|
|||||||||||||||||||||||
►
Sign in to add a comment |
|||||||||||||||||||||||
Comment 1 by jialiul@chromium.org
, Mar 24 2017