AES ARM32 and Thumb2 ASM: fixup ARM32 and add Thumb2#6731
AES ARM32 and Thumb2 ASM: fixup ARM32 and add Thumb2#6731JacobBarthelmeh merged 3 commits intowolfSSL:masterfrom
Conversation
|
|
Detected by the nightly on
Must be from 36b92a4. In this PR, rebased on current master, the errors are: The config for |
|
AES works for me cross-compiling for ARMv7a. |
b1f8a22 to
d786674
Compare
|
for all ARM variants, |
|
retried with Maybe |
|
Upgraded to qemu 8.1.0 here this morning, and reran under that. Same result. Bear in mind these tests were all succeeding until 36b92a4. |
8361ee8 to
1690e1f
Compare
Fix which functions and data are compiled in depending on defines. Better handing of constants. Also fix Aarch64 ed25519 inline assembly.
62e51f6 to
a0ba5d8
Compare
a0ba5d8 to
136cb98
Compare
…urve25519BasePoint.
|
Jenkins, retest this please. |
1 similar comment
|
Jenkins, retest this please. |
|
I have a static analysis finding in CodeSonar that popped up due to this PR. Screenshot attached. I have to admit, I do not understand the logic. line 1015 in wolfcrypt/src/ed25519.c passes in in+1+ED25519_PUB_KEY_SIZE. Problem is that in points to a buffer of ED25519_PUB_KEY_SIZE, so by definition that pointer is outside of the buffer and then used by XMEMCPY. SARIF also attached, this you can load into VS Code to see the flow. |
|
Thank you for letting us know about this so quickly! Let us know if it is not fixed in the PR. Thanks, |

Description
Fix which functions and data are compiled in depending on defines.
Better handing of constants.
Testing
Regression tested ARM32 and Thumb2 assembly including inline.
Checklist