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
At the moment, most *at syscalls will still try to resolve the directory custody before checking the path (with the latter happening in VirtualFileSystem::resolve_path_without_veil), so passing an invalid or non-directory file descriptor (such as 0, 1 or 2) will cause errors that shouldn't happen according to POSIX.
#23099 papered over this for the utimensat functionality in particular, but ideally there should be a structurally sound fix that we can easily apply to all similar syscalls. One potential solution is to not pass the resolved Custody down to resolve_path_without_veil, but rather a fallible lambda function that resolves the Custody on demand if required.
The text was updated successfully, but these errors were encountered:
At the moment, most
*at
syscalls will still try to resolve the directory custody before checking the path (with the latter happening inVirtualFileSystem::resolve_path_without_veil
), so passing an invalid or non-directory file descriptor (such as0
,1
or2
) will cause errors that shouldn't happen according to POSIX.#23099 papered over this for the
utimensat
functionality in particular, but ideally there should be a structurally sound fix that we can easily apply to all similar syscalls. One potential solution is to not pass the resolvedCustody
down toresolve_path_without_veil
, but rather a fallible lambda function that resolves theCustody
on demand if required.The text was updated successfully, but these errors were encountered: