Implement DLQ_GetPtrAuthMask using info from debugserver. … #4110
Conversation
adrian-prantl
requested review from
augusto2112,
JDevlieghere,
jasonmolenda,
kastiglione and
fredriss
|
I assume you're planning to land the generic part of this upstream? If so it might be nice to split this into 2 commits and then cherrypick the one to llvm.org once it's merged here. |
adrian-prantl
force-pushed
the
90718091
branch
2 times, most recently
from
49a0063 to
23fa012
Compare
|
@swift-ci test |
This is needed by the Swift Plugin. See also swiftlang#4110. Differential Revision: https://reviews.llvm.org/D122347
rdar://90718091
|
@swift-ci test |
| return false; | ||
| case DLQ_GetPtrAuthMask: { | ||
| assert(m_process.GetCodeAddressMask() == m_process.GetDataAddressMask() && | ||
| "not supported"); |
There was a problem hiding this comment.
Nit: different code and address masks are not supported?
There was a problem hiding this comment.
Does this code know if the mask returned is used for data or for code addresses? Linux standardized on the code/data masks, but I don't think the distinction is possible in AArch64, I'm not sure what target motivated that choice. I think a sanity check of ensuring that they're the same, if we don't know what this mask might be used for, is correct.
There was a problem hiding this comment.
The swift::reflection::MemoryReader API doesn't distinguish between the two.
|
@swift-ci please test macos |
This is needed by the Swift Plugin. See also swiftlang/llvm-project#4110. Differential Revision: https://reviews.llvm.org/D122347
rdar://90718091