update-engine failing with ASAN error in amd64-generic-asan builder... |
||
Issue descriptionStarting around the middle of July 25, the amd64-generic-asan builder has been consistently failing its UnitTest phase with ASAN errors in the update-engine package. The asan builder logs can be found here: https://cros-goldeneye.corp.google.com/chromeos/legoland/builderHistory?buildConfig=amd64-generic-tot-asan-informational&buildBranch=master The build that started July 24 at 23:15 passes the update-engine package (it failed for a different reason); the next few builds failed in BuildPackages; the next build that made it to UnitTest started July 25 at 15:11. That build, and every one since, has failed with an ASAN failure in update-engine. Here's an example log of the ASAN failure: https://luci-logdog.appspot.com/v/?s=chromeos/buildbucket/cr-buildbucket.appspot.com/8939991820964282800/+/steps/UnitTest/0/stdout
,
Jul 31
Ping? Any progress on this?
,
Aug 2
The following revision refers to this bug: https://chromium.googlesource.com/aosp/platform/system/update_engine/+/68512d42cc9e750f3d6690c27cbd13740a8af142 commit 68512d42cc9e750f3d6690c27cbd13740a8af142 Author: Amin Hassani <ahassani@chromium.org> Date: Thu Aug 02 20:52:54 2018 update_engine: Fix a memory leak in UpdateDeferredByPolicyTest BUG= chromium:868520 TEST=unittests Change-Id: I51026ff3b16bd1f3609248a1afdec95dfed1db3e Reviewed-on: https://chromium-review.googlesource.com/1157923 Commit-Ready: Amin Hassani <ahassani@chromium.org> Tested-by: Amin Hassani <ahassani@chromium.org> Reviewed-by: Sen Jiang <senj@chromium.org> [modify] https://crrev.com/68512d42cc9e750f3d6690c27cbd13740a8af142/update_attempter_unittest.cc
,
Aug 2
The following revision refers to this bug: https://chromium.googlesource.com/aosp/platform/system/update_engine/+/0cd9d7785f1cd56116b0f536e126481e9a5e5374 commit 0cd9d7785f1cd56116b0f536e126481e9a5e5374 Author: Amin Hassani <ahassani@chromium.org> Date: Thu Aug 02 20:52:55 2018 update_engine: ReceivedBytes returns boolean on transfer completion/termination Currently, there are situations that a HttpFetcherDelegate.ReceivedBytes can cause an eventual transfer completion or termination. This can eventually cause the object (holding an instance) of HttpFetcherDelegate to be deallocated. So there should not be any access to any member variable if the object is deallocated. In this CL we add a return value to ReceivedBytes to indicate explicitly that this situation happened so we can be careful about acceessing member variables after a call to this. BUG= chromium:868520 TEST=unittests Change-Id: I18db33910f6171c4e426d9bbe604fa1ae07a56dc Reviewed-on: https://chromium-review.googlesource.com/1158124 Commit-Ready: Amin Hassani <ahassani@chromium.org> Tested-by: Amin Hassani <ahassani@chromium.org> Reviewed-by: Amin Hassani <ahassani@chromium.org> [modify] https://crrev.com/0cd9d7785f1cd56116b0f536e126481e9a5e5374/payload_consumer/download_action.cc [modify] https://crrev.com/0cd9d7785f1cd56116b0f536e126481e9a5e5374/omaha_request_action.cc [modify] https://crrev.com/0cd9d7785f1cd56116b0f536e126481e9a5e5374/common/multi_range_http_fetcher.h [modify] https://crrev.com/0cd9d7785f1cd56116b0f536e126481e9a5e5374/payload_consumer/download_action.h [modify] https://crrev.com/0cd9d7785f1cd56116b0f536e126481e9a5e5374/common/file_fetcher.cc [modify] https://crrev.com/0cd9d7785f1cd56116b0f536e126481e9a5e5374/common/multi_range_http_fetcher.cc [modify] https://crrev.com/0cd9d7785f1cd56116b0f536e126481e9a5e5374/common/http_fetcher.h [modify] https://crrev.com/0cd9d7785f1cd56116b0f536e126481e9a5e5374/common/http_fetcher_unittest.cc [modify] https://crrev.com/0cd9d7785f1cd56116b0f536e126481e9a5e5374/libcurl_http_fetcher.cc [modify] https://crrev.com/0cd9d7785f1cd56116b0f536e126481e9a5e5374/omaha_request_action.h
,
Aug 2
This should be fixed once a new asan builder starts. Feel free to reopen it otherwise. Thanks. |
||
►
Sign in to add a comment |
||
Comment 1 by ahass...@chromium.org
, Jul 27Status: Assigned (was: Untriaged)