https://sourceware.org/bugzilla/show_bug.cgi?id=17896
Cary Coutant <ccoutant at google dot com> changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|NEW |RESOLVED
Blocks|15660 |
Resolution|--- |INVALID
--- Comment #9 from Cary Coutant <ccoutant at google dot com> ---
The leakage in PR 15660 is due to a failure to unlock the plugin object when a
member of a thin archive is claimed by the plugin.
The release_input_file API does in fact release the descriptor. Gold does not
close descriptors right away -- when it releases a descriptor, it merely
becomes available for closing if we run into descriptor pressure later on. As
far as this PR is concerned, gold is working as intended.
In addition, the release_input_file API was only added to go with the
get_input_file API, for use during the all_symbols_read callback. The file
descriptor passed to the claim_file callback is managed by the linker and is
released automatically when the claim_file handler returns (except for the bug
in PR 15660).
The LTO plugin should *not* use release_input_file during the claim_file
handler.
--
You are receiving this mail because:
You are on the CC list for the bug.
_______________________________________________
bug-binutils mailing list
bug-binutils@gnu.org
https://lists.gnu.org/mailman/listinfo/bug-binutils
