Fix sparse-checkout crash due to custom virtual filesystem patches#639
Conversation
|
@dscho @jeffhostetler: The osx-(clang|gcc) test failures seem pretty consistent across three rebuilds, but also unrelated to my changes. Any idea what's going on there? |
I think #640 will fix this. |
|
@dscho @jeffhostetler it does look like at least one of the CI/osx-clang tests remain failing even after #640 was merged this morning. Any thoughts about this? |
This avoids crashes when not using the virtualfilesystem feature. Signed-off-by: Derrick Stolee <stolee@gmail.com> Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
894122f to
6aabf8c
Compare
I think it simply did not synchronize with the newest |
This fixup! commit updates an old virtual filesystem patch to only check its custom logic if the virtual filesystem is enabled. When not enabled, the call to
resolve_dtype()could cause a sparse-index expansion (ensure_full_index()) during an existing scan of the index, leading to use-after-free violations.I verified that my local version with this fix avoids the crash in the case of the Office monorepo.