[SYCL][libdevice] Get rid of builtins for memcpy memset in libdevice #4919
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
jinge90
changed the title
|
The patch is done but we need more extensive testing for it, we are working on enhance tests for libdevice memcpy/set using USM: intel/llvm-test-suite#558 Thanks very much. |
Previously, memcpy and memset are implemented in libdevice by calling
to __builtin_memcpy and __builtin_memset. Such code will lead to
infinite loop if underying CPU runtime implements memcpy/set builtin by
calling memcpy, memset. The function call chain is following:
In libdevice:
memcpy(dest, src, n) {
return __devicelib_memcpy(dest, src, n);
}
__devicelib_memcpy(dest, src, n) {
return __builtin_memcpy(dest, src, n);
}
In CPU runtime:
Handing of __builtin_memcpy<--------------------|
| |
| |
|---->memcpy------>__devicelib_memcpy
In order to fix this, we have to provide implementation for memcpy/set
without using __builtin_*.
|
/summary:run |
jinge90
changed the title
|
Hi, @vzakhari |
|
/summary:run |
bader
reviewed
Nov 17, 2021
Signed-off-by: jinge90 <[email protected]>
bader
approved these changes
Nov 21, 2021
|
Hi, @vzakhari |
vzakhari
approved these changes
Nov 23, 2021
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Previously, memcpy and memset are implemented in libdevice by calling
to __builtin_memcpy and __builtin_memset. Such code will lead to
infinite loop if underying CPU runtime implements memcpy/set builtin by
calling memcpy, memset. The function call chain is following:
In libdevice:
memcpy(dest, src, n) {
return __devicelib_memcpy(dest, src, n);
}
__devicelib_memcpy(dest, src, n) {
return __builtin_memcpy(dest, src, n);
}
In CPU runtime:
Handing of __builtin_memcpy<-------------------|
| |
| |
|---->memcpy------>__devicelib_memcpy
In order to fix this, we have to provide implementation for memcpy/set
without using _builtin*.