On 10/03/2022 19:29, Paul Eggert wrote:
On 3/10/22 05:46, Pádraig Brady wrote:
After looking at the kernel code, it seems that:
fstatat() did _not_ imply AT_NO_AUTOMOUNT from 2.6.38 -> 4.11
I'm not sure it even honored the AT_NO_AUTOMOUNT flag before 4.11
fstatat() did imply AT_NO_AUTOMOUNT since 4.11
Ouch, so this whole thing has been a false alarm? Well, in some sense
that's a relief; in another sense I wonder whether we should undo some
of the recent Gnulib changes.
The changes are a net improvement I think since fewer interfaces are used.
I would remove the AT_NO_AUTOMOUNT parameters to fstatat() though,
since they're redundant it seems, and would only result in confusion
if the patch is applied to remove that flag from the fstatat(2) man page.
thanks,
Pádraig
