You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I have no clue about HWAsan. I guess it's an address-sanitizer and it crashes on the line mentioned in the other issue (https://github.com/python-cffi/cffi/blob/v1.16.0/src/c/_cffi_backend.c#L8038). My only clue would be that it complains that NULL might in theory be different from all-zeroes, which is the initial value of the pointer. While that's correct in theory, and could be fixed here, I also think it's kinda pointless. On all architectures, NULL is the all-zero value, since decades ago.
It is initialized. The C standard says it is initialized to all-zeroes. The part outside the C standard is that all-zeroes happens to be the NULL value on all platforms since forever.
Activity
n-bes commentedon Jul 9, 2024
Catch without dependencies
Dockerfile:
arigo commentedon Jul 9, 2024
I have no clue about HWAsan. I guess it's an address-sanitizer and it crashes on the line mentioned in the other issue (https://github.com/python-cffi/cffi/blob/v1.16.0/src/c/_cffi_backend.c#L8038). My only clue would be that it complains that NULL might in theory be different from all-zeroes, which is the initial value of the pointer. While that's correct in theory, and could be fixed here, I also think it's kinda pointless. On all architectures, NULL is the all-zero value, since decades ago.
n-bes commentedon Jul 9, 2024
I think that
unique_cacheis simply not initialized:cffi/src/c/_cffi_backend.c
Line 388 in ba44abd
Yes, but hardware implementation (arm-mte).
alex commentedon Jul 9, 2024
arigo commentedon Jul 9, 2024
It is initialized. The C standard says it is initialized to all-zeroes. The part outside the C standard is that all-zeroes happens to be the NULL value on all platforms since forever.
arigo commentedon Jul 9, 2024
Sorry if that wasn't clear: try to initialize the value to NULL explicitly:
It shouldn't change anything at all, except make the address sanitizer happy?
n-bes commentedon Jul 10, 2024
Nothing changed at all:
arigo commentedon Jul 10, 2024
Sorry, no idea. Try to look for help in places related to HWAsan.
alex commentedon Jul 12, 2024
This looks an awful lot like google/sanitizers#1245, which seems to suggest this may be an hwasan bug.
n-bes commentedon Aug 16, 2024
@alex, good link. Stack-trace looks like a false-positive, I'll rebuild with additional flags.