[SYCL] Reland community changes to support RawAddress #13358
Conversation
premanandrao
temporarily deployed
to
WindowsCILock
GitHub Actions
Inactive
premanandrao
temporarily deployed
to
WindowsCILock
GitHub Actions
Inactive
|
@intel/dpcpp-cfe-reviewers, can I please get a review on this? You only need to review the minor changes in my second commit. The first commit is a community cherry-pick. |
|
Can you explain why the additional fixes were needed? |
Not sure what you are asking, but the getPointer member function was removed from the Address class. |
I did not realize the code you modified was not present in community. I was wondering why the original PR did not make these changes. |
|
@intel/llvm-gatekeepers, can this be merged please? The test failure about code formatting points to community changes not the additional changes that I made. |
|
Can you please manually squash the commits so author attribution is correct?
Since the bulk of the work here is by @ahatanak |
premanandrao
force-pushed
the
remote_rawaddress
branch
from
7233de0 to
ce1c9ca
Compare
…needed to authenticate signed pointers (#86923) To authenticate pointers, CodeGen needs access to the key and discriminators that were used to sign the pointer. That information is sometimes known from the context, but not always, which is why `Address` needs to hold that information. This patch adds methods and data members to `Address`, which will be needed in subsequent patches to authenticate signed pointers, and uses the newly added methods throughout CodeGen. Although this patch isn't strictly NFC as it causes CodeGen to use different code paths in some cases (e.g., `mergeAddressesInConditionalExpr`), it doesn't cause any changes in functionality as it doesn't add any information needed for authentication. In addition to the changes mentioned above, this patch introduces class `RawAddress`, which contains a pointer that we know is unsigned, and adds several new functions for creating `Address` and `LValue` objects. This reapplies d9a685a, which was reverted because it broke ubsan bots. There seems to be a bug in coroutine code-gen, which is causing EmitTypeCheck to use the wrong alignment. For now, pass alignment zero to EmitTypeCheck so that it can compute the correct alignment based on the passed type (see function EmitCXXMemberOrOperatorMemberCallExpr). Additional fixes needed
premanandrao
temporarily deployed
to
WindowsCILock
GitHub Actions
Inactive
Thank you for pointing this out. Should be fixed now. (Unless I messed it up.) |
4563116 - the author attribution was correct before the manual squash. As far as I can tell it's preserved by GitHub squash and merge, so manual squash is not needed. @ldrumm, did I miss anything? |
premanandrao
temporarily deployed
to
WindowsCILock
GitHub Actions
Inactive
|
@intel/llvm-gatekeepers, will it help if I commit this in as two separate patches? One: that is exclusively the community changes and two: my changes? For a brief period in between, our builds will break, but my changes will fix that. That will keep the authorship entirely separate. (There is further downstream work that I need to do after this is merged here. Hence the time sensitivity.) |
|
No.Even with your manual squash (which I thought would work) there's apparently no way to preserve correct authorship, nor the original commit message. We need an admin to merge this manually rather than github do it. Our policy is broken in the face of cherry picks or similar |
|
@stdale-intel Can you merge this patch manually with the complete, correctly attributed 02b4109, please? |
As it was a complex set of changes
llvm/llvm-project@84780af
was not applied during normal pulldown. This PR relands that commit as well as a couple of
additional fixes needed.