... or otherwise we get an ICE.
Details
Diff Detail
- Repository
- rL LLVM
Event Timeline
clang/lib/Driver/ToolChains/Clang.cpp | ||
---|---|---|
1561 ↗ | (On Diff #212558) | Please move back Driver closer to the first use |
1624 ↗ | (On Diff #212558) | maybe better place is somewhere near llvm/tools/clang/lib/Driver/SanitizerArgs.cpp:1020 |
clang/test/Driver/fsanitize.c | ||
194 ↗ | (On Diff #212558) | Can you also add command line where driver accepts flags? |
clang/lib/Driver/ToolChains/Clang.cpp | ||
---|---|---|
1624 ↗ | (On Diff #212558) | The check is target specific, I prefer to keep it in a function that's already target specific anyway. |
clang/lib/Driver/ToolChains/Clang.cpp | ||
---|---|---|
1632 ↗ | (On Diff #213597) | I've already fixed the line length here. |
clang/lib/Driver/ToolChains/Clang.cpp | ||
---|---|---|
1624 ↗ | (On Diff #212558) | My thoughts here are that the check is both sanitizer specific and platform specific. However it's going to be always sanitizers specific, but in theory we may get more platforms supporting "mte". Also right now err_stack_tagging_requires_hardware_feature is very platform specific. '-fsanitize=memtag' requires hardware support (+memtag) |
clang/lib/Driver/SanitizerArgs.cpp | ||
---|---|---|
1042 ↗ | (On Diff #214185) | I'm a bit worried that this check can be satisfied by, say, a source file or a relative include path of "+mte". |
clang/lib/Driver/SanitizerArgs.cpp | ||
---|---|---|
1042 ↗ | (On Diff #214185) | Fair enough, I've updated the patch. |