win_clang_dbg fails due to ANGLE roll |
||
Issue descriptionThe win_clang_dbg try-bots are failing due to an ANGLE roll landing in https://codereview.chromium.org/2589343002 (in spite of the win_clang_bot failing in the CL for that roll ;): FAILED: obj/third_party/angle/src/vulkan_support/vulkan_loader/loader.obj E:\b\c\cipd\goma/gomacc.exe ../../third_party/llvm-build/Release+Asserts/bin/clang-cl.exe /nologo /showIncludes /FC @obj/third_party/angle/src/vulkan_support/vulkan_loader/loader.obj.rsp /c ../../third_party/vulkan-validation-layers/src/loader/loader.c /Foobj/third_party/angle/src/vulkan_support/vulkan_loader/loader.obj /Fd"obj/third_party/angle/src/vulkan_support/vulkan_loader_c.pdb" In file included from ../../third_party/vulkan-validation-layers/src/loader/loader.c:42: E:\b\c\b\win_clang\src\third_party\vulkan-validation-layers\src\loader\table_ops.h(50,46): error: cast between incompatible calling conventions 'cdecl' and 'stdcall'; calls through this pointer may abort at runtime [-Werror,-Wcast-calling-convention] dev_table->ext_dispatch.dev_ext[i] = (PFN_vkDevExt)vkDevExtError; ^~~~~~~~~~~~~~~~~~~~~~~~~~~ E:\b\c\b\win_clang\src\third_party\vulkan-validation-layers\src\loader\table_ops.h(32,17): note: consider defining 'vkDevExtError' with the 'stdcall' calling convention static VkResult vkDevExtError(VkDevice dev) { ^ __RPC_CALLEE 1 error generated. [5303/35845] LIB obj/third_party/angle/translator.lib [5304/35845] CC obj/third_party/angle/src/vulkan_support/vulkan_loader/trampoline.obj FAILED: obj/third_party/angle/src/vulkan_support/vulkan_loader/trampoline.obj E:\b\c\cipd\goma/gomacc.exe ../../third_party/llvm-build/Release+Asserts/bin/clang-cl.exe /nologo /showIncludes /FC @obj/third_party/angle/src/vulkan_support/vulkan_loader/trampoline.obj.rsp /c ../../third_party/vulkan-validation-layers/src/loader/trampoline.c /Foobj/third_party/angle/src/vulkan_support/vulkan_loader/trampoline.obj /Fd"obj/third_party/angle/src/vulkan_support/vulkan_loader_c.pdb" In file included from ../../third_party/vulkan-validation-layers/src/loader/trampoline.c:35: E:\b\c\b\win_clang\src\third_party\vulkan-validation-layers\src\loader\table_ops.h(50,46): error: cast between incompatible calling conventions 'cdecl' and 'stdcall'; calls through this pointer may abort at runtime [-Werror,-Wcast-calling-convention] dev_table->ext_dispatch.dev_ext[i] = (PFN_vkDevExt)vkDevExtError; ^~~~~~~~~~~~~~~~~~~~~~~~~~~ E:\b\c\b\win_clang\src\third_party\vulkan-validation-layers\src\loader\table_ops.h(32,17): note: consider defining 'vkDevExtError' with the 'stdcall' calling convention static VkResult vkDevExtError(VkDevice dev) { ^ __RPC_CALLEE 1 error generated. Note that the roll also caused several new dependencies to be pulled in under third_party/, which should either be added to .gitignore, or moved to an ANGLE-owned dependencies directory.
,
Dec 21 2016
win_clang_dbg is not part of the default set of CQ bots, nor is it listed in CQ_INCLUDE_TRYBOTS. I'm guessing someone triggered it manually? I'm not sure what the question is here. Are you wondering why the failure was ignored, or what the difference in configs is, or something else?
,
Dec 21 2016
Wez maybe you can answer here - I think the concern in there's a gap in CQ test coverage for Clang 32-bits building. The chromium fyi bots might break randomly. (Maybe that's OK?)
,
Dec 21 2016
IIUC the issue is that vulkan-validation-layers has a legitimate bug in the 32-bit build, but our try-bot coverage seems not to show that up; the only bot configured to build 32-bit seems to be win_clang_dbg. So if the issue is a general 32-bit problem, not Clang specific, then the fact the roll landed suggests we're missing some necessary coverage.
,
Dec 21 2016
We don't cover every possible builder that is on the waterfall in the CQ; some of them are too slow and it would take a lot of resources for very little additional benefit. So we expect some breakages to get through. In particular, win_clang_dbg matches one of the FYI builders, and we generally don't cover the FYI builders, because, well, they're FYI builders, and we expect the FYI builder maintainers to normally fix things rather than revert changes.
,
Dec 21 2016
The following revision refers to this bug: https://chromium.googlesource.com/angle/angle/+/05ac622527441bff0169ac4251939d4ca4ea7d21 commit 05ac622527441bff0169ac4251939d4ca4ea7d21 Author: Jamie Madill <jmadill@chromium.org> Date: Wed Dec 21 19:32:08 2016 Roll Vulkan validation layers. Includes a fix for Clang call cast conversion warning. BUG= chromium:676217 Change-Id: I1b8bf8698714f4280aea52415ae5f96272a2219a Reviewed-on: https://chromium-review.googlesource.com/422440 Commit-Queue: Jamie Madill <jmadill@chromium.org> Reviewed-by: Jamie Madill <jmadill@chromium.org> [modify] https://crrev.com/05ac622527441bff0169ac4251939d4ca4ea7d21/DEPS.chromium [modify] https://crrev.com/05ac622527441bff0169ac4251939d4ca4ea7d21/DEPS
,
Dec 22 2016
The following revision refers to this bug: https://chromium.googlesource.com/chromium/src.git/+/406e37b38ab86429ce67bd26b37f959a38a0f0d4 commit 406e37b38ab86429ce67bd26b37f959a38a0f0d4 Author: cwallez <cwallez@chromium.org> Date: Thu Dec 22 14:53:03 2016 Roll ANGLE d901983..b8a580a https://chromium.googlesource.com/angle/angle.git/+log/d901983..b8a580a BUG= 676333 ,chromium:675997, chromium:676217 ,None,chromium:675819 TBR=geofflang@chromium.org TEST=bots CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.win:win_optional_gpu_tests_rel;master.tryserver.chromium.mac:mac_optional_gpu_tests_rel;master.tryserver.chromium.linux:linux_optional_gpu_tests_rel;master.tryserver.chromium.android:android_optional_gpu_tests_rel Review-Url: https://codereview.chromium.org/2592213002 Cr-Commit-Position: refs/heads/master@{#440415} [modify] https://crrev.com/406e37b38ab86429ce67bd26b37f959a38a0f0d4/DEPS
,
Dec 23 2016
Bot has recovered now. |
||
►
Sign in to add a comment |
||
Comment 1 by jmad...@chromium.org
, Dec 21 2016Components: Internals>GPU>ANGLE
Owner: jmad...@chromium.org
Status: Assigned (was: Untriaged)