Re: cpio -m does not preserve timestamps for symlinks

[Carl Edquist]

Hello Pavel,

> Patch is also proposed here:
> https://www.mail-archive.com/address@hidden/msg00605.html

Thanks for the link; it looks like you've made several attempts to get 
attention about this issue here and downstream, without any luck :-\

Looks like you and I went about it a little differently, but at the end of 
the day they both boil down to calls to utimensat with 
AT_SYMLINK_NOFOLLOW.

I tried to make my patch minimal, changing only the few necessary lines, 
which I hope is easy for the maintainers to review.

Thanks..!
Carl

On Tue, 4 Feb 2020, Pavel Raiskup wrote:

> On Tuesday, February 4, 2020 7:05:21 PM CET Carl Edquist wrote:
> > Hi all,
> >
> > > It seems that for a cpio archive that contains symlinks, the timestamps
> > > for those symlinks are not getting preserved, even with the '-m' option.
> >
> > I believe I found the issue -
> >
> > in copyin.c, copyin_regular_file() and copyin_device() both make calls to
> > set file times (if '-m' was specified), but copyin_link() does not.
> >
> > I was able to fix it by adding a call to set_file_times() in copyin_link()
> > if retain_time_flag is set, just like is done in copyin_device(), and
> > additionally adding the AT_SYMLINK_NOFOLLOW flag to the utimensat() call
> > in fdutimens(), which does the right thing for symlinks and does not
> > change the behavior for non-symlinks.
> >
> > Here is a patch of what I used to get it working:
> >
> >         https://github.com/edquist/cpio/commit/c2ccb84d5e.patch
> >
> > (Patch text also pasted below.)
> >
> > Would you maintainers consider applying this enhancement/fix?
> >
> > Thanks..!
> > Carl
>
> Patch is also proposed here:
> https://www.mail-archive.com/address@hidden/msg00605.html
>
> Pavel