After upgrade to 0.22: atf-check: ERROR: Caught unexpected error: vector (sometimes)
#76
Comments
|
|
|
Here is a FreeBSD example: |
|
Oh, interesting... it's a UBSAN run, so compiler-rt is likely outputting some error message which isn't getting cleanly relayed to the consumer. |
This also has UBSAN enabled. Hmmm... I suspect there's something funky with the new release with CC: @bapt |
atf-check: ERROR: Caught unexpected error: vector message is sometimes seen with UBSAN enabled
atf-check: ERROR: Caught unexpected error: vector message is sometimes seen with UBSAN enabledatf-check: ERROR: Caught unexpected error: vector message is sometimes seen with ASAN/LSAN/UBSAN enabled
|
In the FreeBSD example above, UBSAN is just an environment variable. The code itself is compiled without |
|
Hi Enji, Cheers, |
|
for your information it happen all the time on freebsd with or without libasan, it means the pkg testsuite is broken for weeks, now... |
|
this is the commit that cause the regression: dca2e69 |
I figured it was likely that issue sigh... |
|
My gut feeling is that most of the issues I've addressed on https://github.com/ngie-eign/atf/tree/fix-scan-build-issues will also handle this. I just haven't fixed the issue with |
|
Hi Enji, scanned through your work. I really like thay you are addressing the memory leaks. Looking forward to get that merged and released. |
|
@bapt : do you have a simple repro you can provide to identify the issue seen here? |
|
pkg, ./configure ; make check you will see 3 failures all due to this bug |
To clarify:
? |
atf-check: ERROR: Caught unexpected error: vector message is sometimes seen with ASAN/LSAN/UBSAN enabledatf-check: ERROR: Caught unexpected error: vector (sometimes)
|
Ok, yeah... I see it. |
The previous logic used 2 separate calls to `atf::fs::path::str()` when constructing a `std::vector<char>` to pass to `mkstemp(..)`. This in turn caused grief with how data copying is done in atf-c(3), etc, as the prior code computed the length of the path of an internal buffer in `atf_dynstr` structs. Moreover, the code was manually appending a nul char, which was unnecessary when making the valid assumption that `std::string` is a nul-terminated string. The new code convert the path to an `std::string` once, includes the existing nul char in the buffer, then passes it to mkstemp(3) instead. The code works properly now. Closes: freebsd#76 Signed-off-by: Enji Cooper <ngie@FreeBSD.org>
|
Found the commit that addressed the issue. The problem found by the freebsd/pkg check tests is fixed on HEAD. |
The previous logic used 2 separate calls to `atf::fs::path::str()` when constructing a `std::vector<char>` to pass to `mkstemp(..)`. This in turn caused grief with how data copying is done in atf-c(3), etc, as the prior code computed the length of the path of an internal buffer in `atf_dynstr` structs. Moreover, the code was manually appending a nul char, which was unnecessary when making the valid assumption that `std::string` is a nul-terminated string. The new code convert the path to an `std::string` once, includes the existing nul char in the buffer, then passes it to mkstemp(3) instead. The code works properly now. Closes: freebsd#76 Signed-off-by: Enji Cooper <ngie@FreeBSD.org>
This fixes dev-util/pkgconf's tests with atf-0.22. Bug: freebsd/atf#76 Signed-off-by: Sam James <sam@gentoo.org>
The previous logic used 2 separate calls to `atf::fs::path::str()` when constructing a `std::vector<char>` to pass to `mkstemp(..)`. This in turn caused grief with how data copying is done in atf-c(3), etc, as the prior code computed the length of the path of an internal buffer in `atf_dynstr` structs. Moreover, the code was manually appending a nul char, which was unnecessary when making the valid assumption that `std::string` is a nul-terminated string. The new code convert the path to an `std::string` once, includes the existing nul char in the buffer, then passes it to mkstemp(3) instead. The code works properly now. Closes: #76 Signed-off-by: Enji Cooper <ngie@FreeBSD.org>
Hi,
after uprading to 0.22 I have spurious (random)
test errors in the
kyualogs.Seen on both GitHub MacOS runners as well as on SourceHut FreeBSD runners.
E.g.
https://github.com/kevemueller/pkg/actions/runs/12120407123/job/33789027959
How can I get around these?
The text was updated successfully, but these errors were encountered: