mailing list of musl libc
 help / color / Atom feed
* [musl] coreutils cp mishandles error return from lchmod
@ 2020-01-22 14:15 Rich Felker
  2020-01-22 14:34 ` Florian Weimer
  0 siblings, 1 reply; 21+ messages in thread
From: Rich Felker @ 2020-01-22 14:15 UTC (permalink / raw)
  To: bug-coreutils; +Cc: musl

cp (and perhaps other utilities) are treating EOPNOTSUPP from lchmod
as a hard error rather than an indication that the system does not
support modes for symlinks. This recently came up on
https://bugs.gentoo.org/687236#c17 where users are claiming recent
changes to gnulib made the problem go away, but I'm not sure what the
mechanism was, since the underlying problem is still there. Users only
hit the problem on cross-compiling, presumably due to logic in how
gnulib replaces lchmod, and due to gnulib's replacement wrongly
following symlinks (it just #defines it to chmod). gnulib documents
that the caller must check before calling lchmod that the file is not
a symlink, but this is unsafe in the presence of race conditions,

While lchmod is not a standard function, fchmodat with
AT_SYMLINK_NOFOLLOW is, and specifies EOPNOTSUPP for the case where
the system does not support modes on symlinks. musl provides lchmod as
a simple wrapper for this, yielding a version that is safe.

coreutils should be opting to use the system-provided lchmod, which is
safe, and correctly handling error returns (silently treating
EOPNOTSUPP as success) rather than as hard errors.

Rich

^ permalink raw reply	[flat|nested] 21+ messages in thread

* Re: [musl] coreutils cp mishandles error return from lchmod
  2020-01-22 14:15 [musl] coreutils cp mishandles error return from lchmod Rich Felker
@ 2020-01-22 14:34 ` Florian Weimer
  2020-01-22 14:42   ` Rich Felker
  0 siblings, 1 reply; 21+ messages in thread
From: Florian Weimer @ 2020-01-22 14:34 UTC (permalink / raw)
  To: Rich Felker; +Cc: bug-coreutils, musl

* Rich Felker:

> coreutils should be opting to use the system-provided lchmod, which is
> safe, and correctly handling error returns (silently treating
> EOPNOTSUPP as success) rather than as hard errors.

glibc's lchmod always returns ENOSYS (except on Hurd).  I don't know how
lchmod is used in coreutils, but I suspect it is not particularly
useful.

Thanks,
Florian


^ permalink raw reply	[flat|nested] 21+ messages in thread

* Re: [musl] coreutils cp mishandles error return from lchmod
  2020-01-22 14:34 ` Florian Weimer
@ 2020-01-22 14:42   ` Rich Felker
  2020-01-22 15:08     ` Florian Weimer
  0 siblings, 1 reply; 21+ messages in thread
From: Rich Felker @ 2020-01-22 14:42 UTC (permalink / raw)
  To: Florian Weimer; +Cc: bug-coreutils, musl

On Wed, Jan 22, 2020 at 03:34:18PM +0100, Florian Weimer wrote:
> * Rich Felker:
> 
> > coreutils should be opting to use the system-provided lchmod, which is
> > safe, and correctly handling error returns (silently treating
> > EOPNOTSUPP as success) rather than as hard errors.
> 
> glibc's lchmod always returns ENOSYS (except on Hurd).  I don't know how
> lchmod is used in coreutils, but I suspect it is not particularly
> useful.

When preserving permissions (cp -p, archive extraction, etc.), you
want lchmod to work correctly just for the purpose of *not* following
the link and thereby unwantedly changing the permissions of the link
target. But, fchmodat with AT_SYMLINK_NOFOLLOW works just as well and
is standard, and that's really what coreutils should be using.

Rich

^ permalink raw reply	[flat|nested] 21+ messages in thread

* Re: [musl] coreutils cp mishandles error return from lchmod
  2020-01-22 14:42   ` Rich Felker
@ 2020-01-22 15:08     ` Florian Weimer
  2020-01-22 15:15       ` Rich Felker
  2020-01-22 21:55       ` bug#39236: " Paul Eggert
  0 siblings, 2 replies; 21+ messages in thread
From: Florian Weimer @ 2020-01-22 15:08 UTC (permalink / raw)
  To: Rich Felker; +Cc: 39236, musl

* Rich Felker:

> On Wed, Jan 22, 2020 at 03:34:18PM +0100, Florian Weimer wrote:
>> * Rich Felker:
>> 
>> > coreutils should be opting to use the system-provided lchmod, which is
>> > safe, and correctly handling error returns (silently treating
>> > EOPNOTSUPP as success) rather than as hard errors.
>> 
>> glibc's lchmod always returns ENOSYS (except on Hurd).  I don't know how
>> lchmod is used in coreutils, but I suspect it is not particularly
>> useful.
>
> When preserving permissions (cp -p, archive extraction, etc.), you
> want lchmod to work correctly just for the purpose of *not* following
> the link and thereby unwantedly changing the permissions of the link
> target. But, fchmodat with AT_SYMLINK_NOFOLLOW works just as well and
> is standard, and that's really what coreutils should be using.

I think you misread what I wrote: lchmod *always* returns ENOSYS.  Even
if the file is not a symbolic link.  Likewise, fchmodat with
AT_SYMLINK_NOFOLLOW *always* returns ENOTSUP.

The reason for this is that the kernel does not provide a suitable
system call to implement this, even though some file systems allow a
mode change for symbolic links.  I think we can do better, although I
should note that each time we implement such emulation in userspace, it
comes back to bite us eventually.

Thanks,
Florian


^ permalink raw reply	[flat|nested] 21+ messages in thread

* Re: [musl] coreutils cp mishandles error return from lchmod
  2020-01-22 15:08     ` Florian Weimer
@ 2020-01-22 15:15       ` Rich Felker
  2020-01-22 15:32         ` Florian Weimer
  2020-01-22 21:55       ` bug#39236: " Paul Eggert
  1 sibling, 1 reply; 21+ messages in thread
From: Rich Felker @ 2020-01-22 15:15 UTC (permalink / raw)
  To: Florian Weimer; +Cc: 39236, musl

On Wed, Jan 22, 2020 at 04:08:26PM +0100, Florian Weimer wrote:
> * Rich Felker:
> 
> > On Wed, Jan 22, 2020 at 03:34:18PM +0100, Florian Weimer wrote:
> >> * Rich Felker:
> >> 
> >> > coreutils should be opting to use the system-provided lchmod, which is
> >> > safe, and correctly handling error returns (silently treating
> >> > EOPNOTSUPP as success) rather than as hard errors.
> >> 
> >> glibc's lchmod always returns ENOSYS (except on Hurd).  I don't know how
> >> lchmod is used in coreutils, but I suspect it is not particularly
> >> useful.
> >
> > When preserving permissions (cp -p, archive extraction, etc.), you
> > want lchmod to work correctly just for the purpose of *not* following
> > the link and thereby unwantedly changing the permissions of the link
> > target. But, fchmodat with AT_SYMLINK_NOFOLLOW works just as well and
> > is standard, and that's really what coreutils should be using.
> 
> I think you misread what I wrote: lchmod *always* returns ENOSYS.  Even
> if the file is not a symbolic link.  Likewise, fchmodat with
> AT_SYMLINK_NOFOLLOW *always* returns ENOTSUP.

Yes, I understood that. I was going into why there should be a real
implementation, but didn't make it clear that that was what I was
doing.

> The reason for this is that the kernel does not provide a suitable
> system call to implement this, even though some file systems allow a
> mode change for symbolic links.  I think we can do better, although I
> should note that each time we implement such emulation in userspace, it
> comes back to bite us eventually.

Emulations in userspace that are approximate, have race conditions,
etc. are bad. Ones that are rigorous are good, though.

Rich

^ permalink raw reply	[flat|nested] 21+ messages in thread

* Re: [musl] coreutils cp mishandles error return from lchmod
  2020-01-22 15:15       ` Rich Felker
@ 2020-01-22 15:32         ` Florian Weimer
  2020-01-22 16:07           ` Rich Felker
  0 siblings, 1 reply; 21+ messages in thread
From: Florian Weimer @ 2020-01-22 15:32 UTC (permalink / raw)
  To: Rich Felker; +Cc: 39236, musl

* Rich Felker:

> On Wed, Jan 22, 2020 at 04:08:26PM +0100, Florian Weimer wrote:
>> * Rich Felker:
>> 
>> > On Wed, Jan 22, 2020 at 03:34:18PM +0100, Florian Weimer wrote:
>> >> * Rich Felker:
>> >> 
>> >> > coreutils should be opting to use the system-provided lchmod, which is
>> >> > safe, and correctly handling error returns (silently treating
>> >> > EOPNOTSUPP as success) rather than as hard errors.
>> >> 
>> >> glibc's lchmod always returns ENOSYS (except on Hurd).  I don't know how
>> >> lchmod is used in coreutils, but I suspect it is not particularly
>> >> useful.
>> >
>> > When preserving permissions (cp -p, archive extraction, etc.), you
>> > want lchmod to work correctly just for the purpose of *not* following
>> > the link and thereby unwantedly changing the permissions of the link
>> > target. But, fchmodat with AT_SYMLINK_NOFOLLOW works just as well and
>> > is standard, and that's really what coreutils should be using.
>> 
>> I think you misread what I wrote: lchmod *always* returns ENOSYS.  Even
>> if the file is not a symbolic link.  Likewise, fchmodat with
>> AT_SYMLINK_NOFOLLOW *always* returns ENOTSUP.
>
> Yes, I understood that. I was going into why there should be a real
> implementation, but didn't make it clear that that was what I was
> doing.

Ah, yes, there should be a real implementation if we can get full
lchmod/AT_SYMLINK_NOFOLLOW behavior on file systems that support it.  If
we can't, I'm not sure if there is a point to it.

>> The reason for this is that the kernel does not provide a suitable
>> system call to implement this, even though some file systems allow a
>> mode change for symbolic links.  I think we can do better, although I
>> should note that each time we implement such emulation in userspace, it
>> comes back to bite us eventually.
>
> Emulations in userspace that are approximate, have race conditions,
> etc. are bad. Ones that are rigorous are good, though.

Is there a reason for the S_ISLNK check in the musl implementation?
With current kernels, chmod on the proc pseudo-file will not traverse
the symbolic link, but I have yet to check if this has always been the
case.

Thanks,
Florian


^ permalink raw reply	[flat|nested] 21+ messages in thread

* Re: [musl] coreutils cp mishandles error return from lchmod
  2020-01-22 15:32         ` Florian Weimer
@ 2020-01-22 16:07           ` Rich Felker
  2020-01-22 16:19             ` Florian Weimer
  0 siblings, 1 reply; 21+ messages in thread
From: Rich Felker @ 2020-01-22 16:07 UTC (permalink / raw)
  To: Florian Weimer; +Cc: 39236, musl

On Wed, Jan 22, 2020 at 04:32:45PM +0100, Florian Weimer wrote:
> * Rich Felker:
> 
> > On Wed, Jan 22, 2020 at 04:08:26PM +0100, Florian Weimer wrote:
> >> * Rich Felker:
> >> 
> >> > On Wed, Jan 22, 2020 at 03:34:18PM +0100, Florian Weimer wrote:
> >> >> * Rich Felker:
> >> >> 
> >> >> > coreutils should be opting to use the system-provided lchmod, which is
> >> >> > safe, and correctly handling error returns (silently treating
> >> >> > EOPNOTSUPP as success) rather than as hard errors.
> >> >> 
> >> >> glibc's lchmod always returns ENOSYS (except on Hurd).  I don't know how
> >> >> lchmod is used in coreutils, but I suspect it is not particularly
> >> >> useful.
> >> >
> >> > When preserving permissions (cp -p, archive extraction, etc.), you
> >> > want lchmod to work correctly just for the purpose of *not* following
> >> > the link and thereby unwantedly changing the permissions of the link
> >> > target. But, fchmodat with AT_SYMLINK_NOFOLLOW works just as well and
> >> > is standard, and that's really what coreutils should be using.
> >> 
> >> I think you misread what I wrote: lchmod *always* returns ENOSYS.  Even
> >> if the file is not a symbolic link.  Likewise, fchmodat with
> >> AT_SYMLINK_NOFOLLOW *always* returns ENOTSUP.
> >
> > Yes, I understood that. I was going into why there should be a real
> > implementation, but didn't make it clear that that was what I was
> > doing.
> 
> Ah, yes, there should be a real implementation if we can get full
> lchmod/AT_SYMLINK_NOFOLLOW behavior on file systems that support it.  If
> we can't, I'm not sure if there is a point to it.

The point is to fail when the target is a symlink, rather than
(erroneously and possibly dangerously) applying the chmod to the link
target. Actually supporting link modes is useless. It's the "not
modifying the target" that's important.

> >> The reason for this is that the kernel does not provide a suitable
> >> system call to implement this, even though some file systems allow a
> >> mode change for symbolic links.  I think we can do better, although I
> >> should note that each time we implement such emulation in userspace, it
> >> comes back to bite us eventually.
> >
> > Emulations in userspace that are approximate, have race conditions,
> > etc. are bad. Ones that are rigorous are good, though.
> 
> Is there a reason for the S_ISLNK check in the musl implementation?
> With current kernels, chmod on the proc pseudo-file will not traverse
> the symbolic link, but I have yet to check if this has always been the
> case.

It's explained in the bz you just replied on,
https://sourceware.org/bugzilla/show_bug.cgi?id=14578

The point of the S_ISLNK check is to fail out early with the ENOTSUPP,
which the caller should treat as "success-like", in the non-racing
condition, without the need to open a fd (which may fail with
ENFILE/EMFILE) and without the need for /proc to be mounted.
Otherwise, a different error will be produced when one of those cases
is hit, and the caller will treat it as a real error.

Rich

^ permalink raw reply	[flat|nested] 21+ messages in thread

* Re: [musl] coreutils cp mishandles error return from lchmod
  2020-01-22 16:07           ` Rich Felker
@ 2020-01-22 16:19             ` Florian Weimer
  2020-01-22 17:15               ` Rich Felker
  0 siblings, 1 reply; 21+ messages in thread
From: Florian Weimer @ 2020-01-22 16:19 UTC (permalink / raw)
  To: Rich Felker; +Cc: 39236, musl

* Rich Felker:

> On Wed, Jan 22, 2020 at 04:32:45PM +0100, Florian Weimer wrote:
>> * Rich Felker:
>> 
>> > On Wed, Jan 22, 2020 at 04:08:26PM +0100, Florian Weimer wrote:
>> >> * Rich Felker:
>> >> 
>> >> > On Wed, Jan 22, 2020 at 03:34:18PM +0100, Florian Weimer wrote:
>> >> >> * Rich Felker:
>> >> >> 
>> >> >> > coreutils should be opting to use the system-provided lchmod, which is
>> >> >> > safe, and correctly handling error returns (silently treating
>> >> >> > EOPNOTSUPP as success) rather than as hard errors.
>> >> >> 
>> >> >> glibc's lchmod always returns ENOSYS (except on Hurd).  I don't know how
>> >> >> lchmod is used in coreutils, but I suspect it is not particularly
>> >> >> useful.
>> >> >
>> >> > When preserving permissions (cp -p, archive extraction, etc.), you
>> >> > want lchmod to work correctly just for the purpose of *not* following
>> >> > the link and thereby unwantedly changing the permissions of the link
>> >> > target. But, fchmodat with AT_SYMLINK_NOFOLLOW works just as well and
>> >> > is standard, and that's really what coreutils should be using.
>> >> 
>> >> I think you misread what I wrote: lchmod *always* returns ENOSYS.  Even
>> >> if the file is not a symbolic link.  Likewise, fchmodat with
>> >> AT_SYMLINK_NOFOLLOW *always* returns ENOTSUP.
>> >
>> > Yes, I understood that. I was going into why there should be a real
>> > implementation, but didn't make it clear that that was what I was
>> > doing.
>> 
>> Ah, yes, there should be a real implementation if we can get full
>> lchmod/AT_SYMLINK_NOFOLLOW behavior on file systems that support it.  If
>> we can't, I'm not sure if there is a point to it.
>
> The point is to fail when the target is a symlink, rather than
> (erroneously and possibly dangerously) applying the chmod to the link
> target. Actually supporting link modes is useless. It's the "not
> modifying the target" that's important.

The kernel supports it on some file systems, though:

$ ls -l /tmp/x
l---------. 1 fweimer fweimer 6 Jan 22 15:27 /tmp/x -> /tmp/x

Although mode 0 curiously does not prevent readlink calls.

> It's explained in the bz you just replied on,
> https://sourceware.org/bugzilla/show_bug.cgi?id=14578
>
> The point of the S_ISLNK check is to fail out early with the ENOTSUPP,
> which the caller should treat as "success-like", in the non-racing
> condition, without the need to open a fd (which may fail with
> ENFILE/EMFILE) and without the need for /proc to be mounted.
> Otherwise, a different error will be produced when one of those cases
> is hit, and the caller will treat it as a real error.

Hmm.  The way I read the musl code, the O_PATH descriptor already
exists.  At this point, you can just chmod the O_PATH descriptor, and
have the kernel report EOPNOTSUPP if the file system does not support
that.

Thanks,
Florian


^ permalink raw reply	[flat|nested] 21+ messages in thread

* Re: [musl] coreutils cp mishandles error return from lchmod
  2020-01-22 16:19             ` Florian Weimer
@ 2020-01-22 17:15               ` Rich Felker
  2020-01-22 20:48                 ` Florian Weimer
  0 siblings, 1 reply; 21+ messages in thread
From: Rich Felker @ 2020-01-22 17:15 UTC (permalink / raw)
  To: Florian Weimer; +Cc: 39236, musl

On Wed, Jan 22, 2020 at 05:19:05PM +0100, Florian Weimer wrote:
> * Rich Felker:
> 
> > On Wed, Jan 22, 2020 at 04:32:45PM +0100, Florian Weimer wrote:
> >> * Rich Felker:
> >> 
> >> > On Wed, Jan 22, 2020 at 04:08:26PM +0100, Florian Weimer wrote:
> >> >> * Rich Felker:
> >> >> 
> >> >> > On Wed, Jan 22, 2020 at 03:34:18PM +0100, Florian Weimer wrote:
> >> >> >> * Rich Felker:
> >> >> >> 
> >> >> >> > coreutils should be opting to use the system-provided lchmod, which is
> >> >> >> > safe, and correctly handling error returns (silently treating
> >> >> >> > EOPNOTSUPP as success) rather than as hard errors.
> >> >> >> 
> >> >> >> glibc's lchmod always returns ENOSYS (except on Hurd).  I don't know how
> >> >> >> lchmod is used in coreutils, but I suspect it is not particularly
> >> >> >> useful.
> >> >> >
> >> >> > When preserving permissions (cp -p, archive extraction, etc.), you
> >> >> > want lchmod to work correctly just for the purpose of *not* following
> >> >> > the link and thereby unwantedly changing the permissions of the link
> >> >> > target. But, fchmodat with AT_SYMLINK_NOFOLLOW works just as well and
> >> >> > is standard, and that's really what coreutils should be using.
> >> >> 
> >> >> I think you misread what I wrote: lchmod *always* returns ENOSYS.  Even
> >> >> if the file is not a symbolic link.  Likewise, fchmodat with
> >> >> AT_SYMLINK_NOFOLLOW *always* returns ENOTSUP.
> >> >
> >> > Yes, I understood that. I was going into why there should be a real
> >> > implementation, but didn't make it clear that that was what I was
> >> > doing.
> >> 
> >> Ah, yes, there should be a real implementation if we can get full
> >> lchmod/AT_SYMLINK_NOFOLLOW behavior on file systems that support it.  If
> >> we can't, I'm not sure if there is a point to it.
> >
> > The point is to fail when the target is a symlink, rather than
> > (erroneously and possibly dangerously) applying the chmod to the link
> > target. Actually supporting link modes is useless. It's the "not
> > modifying the target" that's important.
> 
> The kernel supports it on some file systems, though:
> 
> $ ls -l /tmp/x
> l---------. 1 fweimer fweimer 6 Jan 22 15:27 /tmp/x -> /tmp/x
> 
> Although mode 0 curiously does not prevent readlink calls.
> 
> > It's explained in the bz you just replied on,
> > https://sourceware.org/bugzilla/show_bug.cgi?id=14578
> >
> > The point of the S_ISLNK check is to fail out early with the ENOTSUPP,
> > which the caller should treat as "success-like", in the non-racing
> > condition, without the need to open a fd (which may fail with
> > ENFILE/EMFILE) and without the need for /proc to be mounted.
> > Otherwise, a different error will be produced when one of those cases
> > is hit, and the caller will treat it as a real error.
> 
> Hmm.  The way I read the musl code, the O_PATH descriptor already
> exists.  At this point, you can just chmod the O_PATH descriptor, and
> have the kernel report EOPNOTSUPP if the file system does not support
> that.

Oh, you mean the second one after it's already open? Maybe that's ok.
I was concerned it might follow the link and chmod the target at that
point. I thought you were asking about the ealier check before the
O_PATH open.

Rich

^ permalink raw reply	[flat|nested] 21+ messages in thread

* Re: [musl] coreutils cp mishandles error return from lchmod
  2020-01-22 17:15               ` Rich Felker
@ 2020-01-22 20:48                 ` Florian Weimer
  2020-01-22 20:56                   ` Rich Felker
  0 siblings, 1 reply; 21+ messages in thread
From: Florian Weimer @ 2020-01-22 20:48 UTC (permalink / raw)
  To: Rich Felker; +Cc: 39236, musl

* Rich Felker:

>> Hmm.  The way I read the musl code, the O_PATH descriptor already
>> exists.  At this point, you can just chmod the O_PATH descriptor, and
>> have the kernel report EOPNOTSUPP if the file system does not support
>> that.
>
> Oh, you mean the second one after it's already open? Maybe that's ok.

Yes, that's what I meant.

> I was concerned it might follow the link and chmod the target at that
> point.

In my tests, it works.  I think it's also documented behavior for chown
on these pseudo-files.

I also verified that closing an O_PATH descriptor does not release POSIX
advisory locks for the same file.  But I'm wondering if there's still
something we are missing.

Thanks,
Florian


^ permalink raw reply	[flat|nested] 21+ messages in thread

* Re: [musl] coreutils cp mishandles error return from lchmod
  2020-01-22 20:48                 ` Florian Weimer
@ 2020-01-22 20:56                   ` Rich Felker
  2020-01-22 21:05                     ` Florian Weimer
  0 siblings, 1 reply; 21+ messages in thread
From: Rich Felker @ 2020-01-22 20:56 UTC (permalink / raw)
  To: Florian Weimer; +Cc: 39236, musl

On Wed, Jan 22, 2020 at 09:48:14PM +0100, Florian Weimer wrote:
> * Rich Felker:
> 
> >> Hmm.  The way I read the musl code, the O_PATH descriptor already
> >> exists.  At this point, you can just chmod the O_PATH descriptor, and
> >> have the kernel report EOPNOTSUPP if the file system does not support
> >> that.
> >
> > Oh, you mean the second one after it's already open? Maybe that's ok.
> 
> Yes, that's what I meant.
> 
> > I was concerned it might follow the link and chmod the target at that
> > point.
> 
> In my tests, it works.  I think it's also documented behavior for chown
> on these pseudo-files.

Do you know where we might find that documentation?

> I also verified that closing an O_PATH descriptor does not release POSIX
> advisory locks for the same file.  But I'm wondering if there's still
> something we are missing.

Thanks, I hadn't thought to check that, but wouldn't have expected it
to be a problem since O_PATH is not actually open to the file.

Rich

^ permalink raw reply	[flat|nested] 21+ messages in thread

* Re: [musl] coreutils cp mishandles error return from lchmod
  2020-01-22 20:56                   ` Rich Felker
@ 2020-01-22 21:05                     ` Florian Weimer
  0 siblings, 0 replies; 21+ messages in thread
From: Florian Weimer @ 2020-01-22 21:05 UTC (permalink / raw)
  To: Rich Felker; +Cc: 39236, musl

* Rich Felker:

> On Wed, Jan 22, 2020 at 09:48:14PM +0100, Florian Weimer wrote:
>> * Rich Felker:
>> 
>> >> Hmm.  The way I read the musl code, the O_PATH descriptor already
>> >> exists.  At this point, you can just chmod the O_PATH descriptor, and
>> >> have the kernel report EOPNOTSUPP if the file system does not support
>> >> that.
>> >
>> > Oh, you mean the second one after it's already open? Maybe that's ok.
>> 
>> Yes, that's what I meant.
>> 
>> > I was concerned it might follow the link and chmod the target at that
>> > point.
>> 
>> In my tests, it works.  I think it's also documented behavior for chown
>> on these pseudo-files.
>
> Do you know where we might find that documentation?

Ugh.  I'm probably misremembering.  It may have been a rejection of
patch to implement another f*at system call with AT_EMPTY_PATH support.
(I did find your 2013 message describing the chmod and chown behavior.)

Thanks,
Florian


^ permalink raw reply	[flat|nested] 21+ messages in thread

* Re: bug#39236: [musl] coreutils cp mishandles error return from lchmod
  2020-01-22 15:08     ` Florian Weimer
  2020-01-22 15:15       ` Rich Felker
@ 2020-01-22 21:55       ` " Paul Eggert
  2020-01-22 22:05         ` Rich Felker
  1 sibling, 1 reply; 21+ messages in thread
From: Paul Eggert @ 2020-01-22 21:55 UTC (permalink / raw)
  To: Florian Weimer, Rich Felker; +Cc: musl, 39236

On 1/22/20 7:08 AM, Florian Weimer wrote:
> I think you misread what I wrote: lchmod*always*  returns ENOSYS.  Even
> if the file is not a symbolic link.  Likewise, fchmodat with
> AT_SYMLINK_NOFOLLOW *always* returns ENOTSUP.

That's too bad, because coreutils (and many other applications, I 
expect) assume that lchmod (and fchmodat with AT_SYMLINK_NOFOLLOW) to 
act like chmod except not follow symlinks, in order to make it less 
likely that the application will run afoul of a symlink race and chmod 
the wrong file. Isn't that how the Linux fstatat call behaves? And if 
so, why does glibc fstatat refuse to support this behavior?

To work around this bug, I suppose coreutils etc. should do something 
like the following:

1. Never use lchmod since the porting nightmare is bad enough without it.

2. On non-glibc systems (or glibc systems where the bug is fixed), use 
fchmodat with AT_SYMLINK_NOFOLLOW.

3. On glibc systems with the bug, use openat with AT_SYMLINK_NOFOLLOW 
and O_PATH, and then fchmod the resulting file descriptor.

Does this sound right? Or is there some O_PATH gotcha that I haven't 
thought about?

Come to think of it, perhaps the best thing would be to change Gnulib's 
lchmod and fchmodat modules so that they do what applications expect, 
even on buggy glibc systems. (Which would be ironic, since Gnulib's main 
goal is to put wrappers around other libraries so that they look more 
like glibc.)

^ permalink raw reply	[flat|nested] 21+ messages in thread

* Re: bug#39236: [musl] coreutils cp mishandles error return from lchmod
  2020-01-22 21:55       ` bug#39236: " Paul Eggert
@ 2020-01-22 22:05         ` Rich Felker
  2020-02-08  0:37           ` Paul Eggert
  0 siblings, 1 reply; 21+ messages in thread
From: Rich Felker @ 2020-01-22 22:05 UTC (permalink / raw)
  To: Paul Eggert; +Cc: Florian Weimer, musl, 39236

On Wed, Jan 22, 2020 at 01:55:57PM -0800, Paul Eggert wrote:
> On 1/22/20 7:08 AM, Florian Weimer wrote:
> >I think you misread what I wrote: lchmod*always*  returns ENOSYS.  Even
> >if the file is not a symbolic link.  Likewise, fchmodat with
> >AT_SYMLINK_NOFOLLOW *always* returns ENOTSUP.
> 
> That's too bad, because coreutils (and many other applications, I
> expect) assume that lchmod (and fchmodat with AT_SYMLINK_NOFOLLOW)
> to act like chmod except not follow symlinks, in order to make it
> less likely that the application will run afoul of a symlink race
> and chmod the wrong file. Isn't that how the Linux fstatat call
> behaves? And if so, why does glibc fstatat refuse to support this
> behavior?

I think you're confusing fchmodat with fstatat. The Linux fchmodat
syscall lacks a flags argument and thus doesn't suffice to implement
fchmodat. The fstatat syscall does work.

> To work around this bug, I suppose coreutils etc. should do
> something like the following:
> 
> 1. Never use lchmod since the porting nightmare is bad enough without it.
> 
> 2. On non-glibc systems (or glibc systems where the bug is fixed),
> use fchmodat with AT_SYMLINK_NOFOLLOW.
> 
> 3. On glibc systems with the bug, use openat with
> AT_SYMLINK_NOFOLLOW and O_PATH, and then fchmod the resulting file
> descriptor.
> 
> Does this sound right? Or is there some O_PATH gotcha that I haven't
> thought about?

I think fchmod historically did not work on O_PATH file descriptors,
which is why musl is using chmod on the procfs magic symlink. However,
fchmodat might work too with an empty pathname; I'm not sure.

I think these fixes are better encapsulated as a replacement for
missing/broken fchmodat, rather than putting the logic in individual
utilities or coreutils-specific library code.

Also, note that if you want to skip checking stat to make sure you
didn't open a symlink with O_PATH, that depends on confirming
Florian's claim that the kernel documents it will not follow the
symlink.

> Come to think of it, perhaps the best thing would be to change
> Gnulib's lchmod and fchmodat modules so that they do what
> applications expect, even on buggy glibc systems. (Which would be
> ironic, since Gnulib's main goal is to put wrappers around other
> libraries so that they look more like glibc.)

I think we're approaching a consensus that glibc should fix this too,
so then it would just be gnulib matching the fix.

Rich

^ permalink raw reply	[flat|nested] 21+ messages in thread

* Re: bug#39236: [musl] coreutils cp mishandles error return from lchmod
  2020-01-22 22:05         ` Rich Felker
@ 2020-02-08  0:37           ` Paul Eggert
  2020-02-12 11:50             ` Florian Weimer
  0 siblings, 1 reply; 21+ messages in thread
From: Paul Eggert @ 2020-02-08  0:37 UTC (permalink / raw)
  To: Rich Felker; +Cc: Florian Weimer, musl, 39236, Gnulib bugs

[-- Attachment #1: Type: text/plain, Size: 529 bytes --]

On 1/22/20 2:05 PM, Rich Felker wrote:
> I think we're approaching a consensus that glibc should fix this too,
> so then it would just be gnulib matching the fix.

I installed the attached patch to Gnulib in preparation for the upcoming 
glibc fix. The patch causes fchmodat with AT_SYMLINK_NOFOLLOW to work on 
non-symlinks, and similarly for lchmod on non-symlinks. The idea is to 
avoid this sort of problem in the future, and to let Coreutils etc. work 
on older platforms as if glibc 2.32 (or whatever) is already in place.

[-- Attachment #2: 0001-fchmodat-AT_SYMLINK_NOFOLLOW-fix-for-non-symlinks.patch --]
[-- Type: text/x-patch, Size: 24627 bytes --]

From b16a04394121e7396569a13161dba02c6752b19f Mon Sep 17 00:00:00 2001
From: Paul Eggert <eggert@cs.ucla.edu>
Date: Fri, 7 Feb 2020 16:34:12 -0800
Subject: [PATCH] fchmodat: AT_SYMLINK_NOFOLLOW fix for non-symlinks

Fix lchmod, and fchmodat with AT_SYMLINK_NOFOLLOW, so that
they act like chmod on non-symlinks.
* NEWS:
* doc/glibc-functions/lchmod.texi (lchmod):
* doc/posix-functions/fchmodat.texi (fchmodat):
Mention this.
* lib/fchmodat.c: Define __need_system_sys_stat_h before including
config.h, and undef it after including sys/stat.h the first time.
Include fcntl.h, stdio.h, unistd.h, intprops.h, and include
sys/stat.h a second time after defining orig_fchmodat.
(orig_fchmodat) [HAVE_FCHMODAT]: New function.
(fchmodat) [HAVE_FCHMODAT]: Work around the AT_SYMLINK_NOFOLLOW bug.
* lib/lchmod.c: New file.
* lib/sys_stat.in.h (fchmodat, lchmod):
Support replacing these functions.
* m4/fchmodat.m4 (gl_FUNC_FCHMODAT): If fchmodat exists,
test that AT_SYMLINK_NOFOLLOW works on non-symlinks.
* m4/lchmod.m4 (gl_FUNC_LCHMOD): Check for lstat.
Test that lchmod works on non-symlinks.
* m4/sys_stat_h.m4 (gl_SYS_STAT_H_DEFAULTS):
Default REPLACE_FCHMODAT and REPLACE_LCHMOD to 0.
* modules/fchmodat (Depends-on): Add fstatat, intprops, lchmod, unistd.
(Depends-on, configure.ac): Check REPLACE_FCHMODAT too.
* modules/lchmod (Files): Add lib/lchmod.c.
(Depends-on): Add errno, fcntl-h, fchmodat, intprops, lstat, unistd.
(configure.ac): Compile lchmod.c if needed.
(lib_SOURCES): Add lchmod.c.
* modules/sys_stat (sys/stat.h): Substitute REPLACE_FCHMODAT
and REPLACE_LCHMOD.
* tests/test-fchmodat.c: Include fcntl.h, sys/stat.h.
(main): Test fchmodat with AT_SYMLINK_NOFOLLOW on non-symlinks.
---
 ChangeLog                         | 35 ++++++++++++
 NEWS                              |  7 +++
 doc/glibc-functions/lchmod.texi   |  4 ++
 doc/posix-functions/fchmodat.texi | 11 ++--
 lib/fchmodat.c                    | 89 +++++++++++++++++++++++++------
 lib/lchmod.c                      | 72 +++++++++++++++++++++++++
 lib/sys_stat.in.h                 | 41 +++++++-------
 m4/fchmodat.m4                    | 48 ++++++++++++++++-
 m4/lchmod.m4                      | 52 ++++++++++++++++--
 m4/sys_stat_h.m4                  |  4 +-
 modules/fchmodat                  | 10 ++--
 modules/lchmod                    | 13 ++++-
 modules/sys_stat                  |  2 +
 tests/test-fchmodat.c             | 10 ++++
 14 files changed, 348 insertions(+), 50 deletions(-)
 create mode 100644 lib/lchmod.c

diff --git a/ChangeLog b/ChangeLog
index 99e0c2e9e..71dcaba6c 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,3 +1,38 @@
+2020-02-07  Paul Eggert  <eggert@cs.ucla.edu>
+
+	fchmodat: AT_SYMLINK_NOFOLLOW fix for non-symlinks
+	Fix lchmod, and fchmodat with AT_SYMLINK_NOFOLLOW, so that
+	they act like chmod on non-symlinks.
+	* NEWS:
+	* doc/glibc-functions/lchmod.texi (lchmod):
+	* doc/posix-functions/fchmodat.texi (fchmodat):
+	Mention this.
+	* lib/fchmodat.c: Define __need_system_sys_stat_h before including
+	config.h, and undef it after including sys/stat.h the first time.
+	Include fcntl.h, stdio.h, unistd.h, intprops.h, and include
+	sys/stat.h a second time after defining orig_fchmodat.
+	(orig_fchmodat) [HAVE_FCHMODAT]: New function.
+	(fchmodat) [HAVE_FCHMODAT]: Work around the AT_SYMLINK_NOFOLLOW bug.
+	* lib/lchmod.c: New file.
+	* lib/sys_stat.in.h (fchmodat, lchmod):
+	Support replacing these functions.
+	* m4/fchmodat.m4 (gl_FUNC_FCHMODAT): If fchmodat exists,
+	test that AT_SYMLINK_NOFOLLOW works on non-symlinks.
+	* m4/lchmod.m4 (gl_FUNC_LCHMOD): Check for lstat.
+	Test that lchmod works on non-symlinks.
+	* m4/sys_stat_h.m4 (gl_SYS_STAT_H_DEFAULTS):
+	Default REPLACE_FCHMODAT and REPLACE_LCHMOD to 0.
+	* modules/fchmodat (Depends-on): Add fstatat, intprops, lchmod, unistd.
+	(Depends-on, configure.ac): Check REPLACE_FCHMODAT too.
+	* modules/lchmod (Files): Add lib/lchmod.c.
+	(Depends-on): Add errno, fcntl-h, fchmodat, intprops, lstat, unistd.
+	(configure.ac): Compile lchmod.c if needed.
+	(lib_SOURCES): Add lchmod.c.
+	* modules/sys_stat (sys/stat.h): Substitute REPLACE_FCHMODAT
+	and REPLACE_LCHMOD.
+	* tests/test-fchmodat.c: Include fcntl.h, sys/stat.h.
+	(main): Test fchmodat with AT_SYMLINK_NOFOLLOW on non-symlinks.
+
 2020-02-05  Marc Dionne  <marc.dionne@auristor.com>  (tiny change)
 
 	mountlist: Consider AFS filesystems as remote
diff --git a/NEWS b/NEWS
index dc5cc71f9..bc81dfc28 100644
--- a/NEWS
+++ b/NEWS
@@ -58,6 +58,13 @@ User visible incompatible changes
 
 Date        Modules         Changes
 
+2020-02-07  fchmodat        When applied to non-symlinks, these now act like
+            lchmod          chmod (the BSD behavior, which POSIX requires for
+                            fchmodat + AT_SYMLINK_NOFOLLOW), instead of failing
+                            (the GNU/Linux behavior through glibc 2.31).
+                            Future versions of GNU/Linux are planned to act as
+                            per POSIX and BSD.
+
 2020-01-15  gc-pbkdf2-sha1  This module is deprecated.  Use gc-pbkdf2 instead.
 
 2019-12-12  dfa             Its API now uses ptrdiff_t instead of size_t.
diff --git a/doc/glibc-functions/lchmod.texi b/doc/glibc-functions/lchmod.texi
index aea782dc2..6cc48b453 100644
--- a/doc/glibc-functions/lchmod.texi
+++ b/doc/glibc-functions/lchmod.texi
@@ -9,6 +9,10 @@ Portability problems fixed by Gnulib:
 @item
 This function is missing on some platforms:
 OpenBSD 3.8, Minix 3.1.8, AIX 5.1, IRIX 6.5, Solaris 11.4, Cygwin, mingw, MSVC 14, Android 9.0.
+@item
+This function always fails with @code{errno} set to @code{ENOSYS},
+even when the file is not a symbolic link:
+GNU/Linux with glibc 2.31.
 @end itemize
 
 Portability problems not fixed by Gnulib:
diff --git a/doc/posix-functions/fchmodat.texi b/doc/posix-functions/fchmodat.texi
index 6add5ba89..4d190310e 100644
--- a/doc/posix-functions/fchmodat.texi
+++ b/doc/posix-functions/fchmodat.texi
@@ -9,6 +9,11 @@ Gnulib module: fchmodat
 Portability problems fixed by Gnulib:
 @itemize
 @item
+When given the @code{AT_SYMLINK_NOFOLLOW} flag,
+this function fails with @code{errno} set to @code{ENOTSUP},
+even when the file is not a symbolic link:
+GNU/Linux and Cygwin with glibc 2.31.
+@item
 This function is missing on some platforms:
 glibc 2.3.6, Mac OS X 10.5, FreeBSD 6.0, NetBSD 5.0, OpenBSD 3.8, Minix 3.1.8,
 AIX 5.1, HP-UX 11, IRIX 6.5, Solaris 10, Cygwin 1.5.x, mingw, MSVC 14.
@@ -19,9 +24,5 @@ Portability problems not fixed by Gnulib:
 @itemize
 @item
 Some platforms do not allow changing the access bits on symbolic
-links.  POSIX states that @code{fchmodat(@dots{},AT_SYMLINK_NOFOLLOW)}
-may fail with @code{EOPNOTSUPP} when called on a symlink, but some
-platforms, as well as the gnulib replacement, fail for any use of
-AT_SYMLINK_NOFOLLOW even if the target was not a symlink:
-glibc, Cygwin.
+links.
 @end itemize
diff --git a/lib/fchmodat.c b/lib/fchmodat.c
index 0d4891b3e..c6b8ef768 100644
--- a/lib/fchmodat.c
+++ b/lib/fchmodat.c
@@ -16,29 +16,42 @@
 
 /* written by Jim Meyering */
 
+/* If the user's config.h happens to include <sys/stat.h>, let it include only
+   the system's <sys/stat.h> here, so that orig_fchmodat doesn't recurse to
+   rpl_fchmodat.  */
+#define __need_system_sys_stat_h
 #include <config.h>
 
 /* Specification.  */
 #include <sys/stat.h>
+#undef __need_system_sys_stat_h
 
 #include <errno.h>
+#include <fcntl.h>
+#include <stdio.h>
 #include <stdlib.h>
+#include <unistd.h>
 
-#ifndef HAVE_LCHMOD
-/* Use a different name, to avoid conflicting with any
-   system-supplied declaration.  */
-# undef lchmod
-# define lchmod lchmod_rpl
+#if HAVE_FCHMODAT
 static int
-lchmod (char const *f _GL_UNUSED, mode_t m _GL_UNUSED)
+orig_fchmodat (int dir, char const *file, mode_t mode, int flags)
 {
-  errno = ENOSYS;
-  return -1;
+  return fchmodat (dir, file, mode, flags);
 }
 #endif
 
-/* Solaris 10 has no function like this.
-   Invoke chmod or lchmod on file, FILE, using mode MODE, in the directory
+#ifdef __osf__
+/* Write "sys/stat.h" here, not <sys/stat.h>, otherwise OSF/1 5.1 DTK cc
+   eliminates this include because of the preliminary #include <sys/stat.h>
+   above.  */
+# include "sys/stat.h"
+#else
+# include <sys/stat.h>
+#endif
+
+#include <intprops.h>
+
+/* Invoke chmod or lchmod on FILE, using mode MODE, in the directory
    open on descriptor FD.  If possible, do it without changing the
    working directory.  Otherwise, resort to using save_cwd/fchdir,
    then (chmod|lchmod)/restore_cwd.  If either the save_cwd or the
@@ -46,10 +59,52 @@ lchmod (char const *f _GL_UNUSED, mode_t m _GL_UNUSED)
    Note that an attempt to use a FLAG value of AT_SYMLINK_NOFOLLOW
    on a system without lchmod support causes this function to fail.  */
 
-#define AT_FUNC_NAME fchmodat
-#define AT_FUNC_F1 lchmod
-#define AT_FUNC_F2 chmod
-#define AT_FUNC_USE_F1_COND AT_SYMLINK_NOFOLLOW
-#define AT_FUNC_POST_FILE_PARAM_DECLS , mode_t mode, int flag
-#define AT_FUNC_POST_FILE_ARGS        , mode
-#include "at-func.c"
+#if HAVE_FCHMODAT
+int
+fchmodat (int dir, char const *file, mode_t mode, int flags)
+{
+  if (flags & AT_SYMLINK_NOFOLLOW)
+    {
+# ifdef O_PATH
+      int fd = openat (dir, file, O_PATH | O_NOFOLLOW | O_CLOEXEC);
+      if (fd < 0)
+        return fd;
+      static char const fmt[] = "/proc/self/fd/%d";
+      char buf[sizeof fmt - sizeof "%d" + INT_BUFSIZE_BOUND (int)];
+      sprintf (buf, fmt, fd);
+      int chmod_result = chmod (buf, mode);
+      int chmod_errno = errno;
+      close (fd);
+      if (chmod_result == 0)
+        return chmod_result;
+      if (chmod_errno != ENOENT)
+        {
+          errno = chmod_errno;
+          return chmod_result;
+        }
+      /* /proc is not mounted; fall back on racy implementation.  */
+# endif
+
+      struct stat st;
+      int fstatat_result = fstatat (dir, file, &st, AT_SYMLINK_NOFOLLOW);
+      if (fstatat_result != 0)
+        return fstatat_result;
+      if (S_ISLNK (st.st_mode))
+        {
+          errno = EOPNOTSUPP;
+          return -1;
+        }
+      flags &= ~AT_SYMLINK_NOFOLLOW;
+    }
+
+  return orig_fchmodat (dir, file, mode, flags);
+}
+#else
+# define AT_FUNC_NAME fchmodat
+# define AT_FUNC_F1 lchmod
+# define AT_FUNC_F2 chmod
+# define AT_FUNC_USE_F1_COND AT_SYMLINK_NOFOLLOW
+# define AT_FUNC_POST_FILE_PARAM_DECLS , mode_t mode, int flag
+# define AT_FUNC_POST_FILE_ARGS        , mode
+# include "at-func.c"
+#endif
diff --git a/lib/lchmod.c b/lib/lchmod.c
new file mode 100644
index 000000000..cc260ce4d
--- /dev/null
+++ b/lib/lchmod.c
@@ -0,0 +1,72 @@
+/* Implement lchmod on platforms where it does not work correctly.
+
+   Copyright 2020 Free Software Foundation, Inc.
+
+   This program is free software: you can redistribute it and/or modify
+   it under the terms of the GNU General Public License as published by
+   the Free Software Foundation; either version 3 of the License, or
+   (at your option) any later version.
+
+   This program is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+   GNU General Public License for more details.
+
+   You should have received a copy of the GNU General Public License
+   along with this program.  If not, see <https://www.gnu.org/licenses/>.  */
+
+/* written by Paul Eggert */
+
+#include <config.h>
+
+#include <sys/stat.h>
+
+#include <errno.h>
+#include <fcntl.h>
+#include <stdio.h>
+#include <unistd.h>
+
+#include <intprops.h>
+
+/* Work like lchmod, except when FILE is a symbolic link.
+   In that case, set errno to EOPNOTSUPP and return -1.  */
+
+int
+lchmod (char const *file, mode_t mode)
+{
+#if HAVE_FCHMODAT
+  return fchmodat (AT_FDCWD, file, mode, AT_SYMLINK_NOFOLLOW);
+#elif defined O_PATH && defined AT_FDCWD
+  int fd = openat (AT_FDCWD, file, O_PATH | O_NOFOLLOW | O_CLOEXEC);
+  if (fd < 0)
+    return fd;
+  static char const fmt[] = "/proc/self/fd/%d";
+  char buf[sizeof fmt - sizeof "%d" + INT_BUFSIZE_BOUND (int)];
+  sprintf (buf, fmt, fd);
+  int chmod_result = chmod (buf, mode);
+  int chmod_errno = errno;
+  close (fd);
+  if (chmod_result == 0)
+    return chmod_result;
+  if (chmod_errno != ENOENT)
+    {
+      errno = chmod_errno;
+      return chmod_result;
+    }
+  /* /proc is not mounted; fall back on racy implementation.  */
+#endif
+
+#if HAVE_LSTAT
+  struct stat st;
+  int lstat_result = lstat (file, &st);
+  if (lstat_result != 0)
+    return lstat_result;
+  if (S_ISLNK (st.st_mode))
+    {
+      errno = EOPNOTSUPP;
+      return -1;
+    }
+#endif
+
+  return chmod (file, mode);
+}
diff --git a/lib/sys_stat.in.h b/lib/sys_stat.in.h
index 5c891e7b8..4f9eb5976 100644
--- a/lib/sys_stat.in.h
+++ b/lib/sys_stat.in.h
@@ -392,13 +392,25 @@ struct stat
 
 
 #if @GNULIB_FCHMODAT@
-# if !@HAVE_FCHMODAT@
+# if @REPLACE_FCHMODAT@
+#  if !(defined __cplusplus && defined GNULIB_NAMESPACE)
+#   undef fchmodat
+#   define fchmodat rpl_fchmodat
+#  endif
+_GL_FUNCDECL_RPL (fchmodat, int,
+                  (int fd, char const *file, mode_t mode, int flag)
+                  _GL_ARG_NONNULL ((2)));
+_GL_CXXALIAS_RPL (fchmodat, int,
+                  (int fd, char const *file, mode_t mode, int flag));
+# else
+#  if !@HAVE_FCHMODAT@
 _GL_FUNCDECL_SYS (fchmodat, int,
                   (int fd, char const *file, mode_t mode, int flag)
                   _GL_ARG_NONNULL ((2)));
-# endif
+#  endif
 _GL_CXXALIAS_SYS (fchmodat, int,
                   (int fd, char const *file, mode_t mode, int flag));
+# endif
 _GL_CXXALIASWARN (fchmodat);
 #elif defined GNULIB_POSIXCHECK
 # undef fchmodat
@@ -502,31 +514,24 @@ _GL_WARN_ON_USE (futimens, "futimens is not portable - "
 #if @GNULIB_LCHMOD@
 /* Change the mode of FILENAME to MODE, without dereferencing it if FILENAME
    denotes a symbolic link.  */
-# if !@HAVE_LCHMOD@
-/* The lchmod replacement follows symbolic links.  Callers should take
-   this into account; lchmod should be applied only to arguments that
-   are known to not be symbolic links.  On hosts that lack lchmod,
-   this can lead to race conditions between the check and the
-   invocation of lchmod, but we know of no workarounds that are
-   reliable in general.  You might try requesting support for lchmod
-   from your operating system supplier.  */
+# if @REPLACE_LCHMOD@
 #  if !(defined __cplusplus && defined GNULIB_NAMESPACE)
-#   define lchmod chmod
+#   undef lchmod
+#   define lchmod rpl_lchmod
 #  endif
-/* Need to cast, because on mingw, the second parameter of chmod is
-                                                int mode.  */
-_GL_CXXALIAS_RPL_CAST_1 (lchmod, chmod, int,
-                         (const char *filename, mode_t mode));
+_GL_FUNCDECL_RPL (lchmod, int,
+                  (char const *filename, mode_t mode)
+                  _GL_ARG_NONNULL ((1)));
+_GL_CXXALIAS_RPL (lchmod, int,
+                  (char const *filename, mode_t mode));
 # else
-#  if 0 /* assume already declared */
+#  if !@HAVE_LCHMOD@
 _GL_FUNCDECL_SYS (lchmod, int, (const char *filename, mode_t mode)
                                _GL_ARG_NONNULL ((1)));
 #  endif
 _GL_CXXALIAS_SYS (lchmod, int, (const char *filename, mode_t mode));
 # endif
-# if @HAVE_LCHMOD@
 _GL_CXXALIASWARN (lchmod);
-# endif
 #elif defined GNULIB_POSIXCHECK
 # undef lchmod
 # if HAVE_RAW_DECL_LCHMOD
diff --git a/m4/fchmodat.m4 b/m4/fchmodat.m4
index 460a4dc17..8195ef668 100644
--- a/m4/fchmodat.m4
+++ b/m4/fchmodat.m4
@@ -1,4 +1,4 @@
-# fchmodat.m4 serial 1
+# fchmodat.m4 serial 2
 dnl Copyright (C) 2004-2020 Free Software Foundation, Inc.
 dnl This file is free software; the Free Software Foundation
 dnl gives unlimited permission to copy and/or distribute it,
@@ -13,5 +13,51 @@ AC_DEFUN([gl_FUNC_FCHMODAT],
   AC_CHECK_FUNCS_ONCE([fchmodat lchmod])
   if test $ac_cv_func_fchmodat != yes; then
     HAVE_FCHMODAT=0
+  else
+    AC_CACHE_CHECK(
+      [whether fchmodat+AT_SYMLINK_NOFOLLOW works on non-symlinks],
+      [gl_cv_func_fchmodat_works],
+      [AC_RUN_IFELSE(
+         [AC_LANG_PROGRAM(
+            [
+              AC_INCLUDES_DEFAULT[
+              #include <fcntl.h>
+              #ifndef S_IRUSR
+               #define S_IRUSR 0400
+              #endif
+              #ifndef S_IWUSR
+               #define S_IWUSR 0200
+              #endif
+              #ifndef S_IRWXU
+               #define S_IRWXU 0700
+              #endif
+              #ifndef S_IRWXG
+               #define S_IRWXG 0070
+              #endif
+              #ifndef S_IRWXO
+               #define S_IRWXO 0007
+              #endif
+            ]],
+            [[
+              int permissive = S_IRWXU | S_IRWXG | S_IRWXO;
+              int desired = S_IRUSR | S_IWUSR;
+              static char const f[] = "conftest.fchmodat";
+              struct stat st;
+              if (creat (f, permissive) < 0)
+                return 1;
+              if (fchmodat (AT_FDCWD, f, desired, AT_SYMLINK_NOFOLLOW) != 0)
+                return 1;
+              if (stat (f, &st) != 0)
+                return 1;
+              return ! ((st.st_mode & permissive) == desired);
+            ]])],
+         [gl_cv_func_fchmodat_works=yes],
+         [gl_cv_func_fchmodat_works=no],
+         [gl_cv_func_fchmodat_works=$gl_cross_guess_normal])
+       rm -f conftest.fchmodat])
+    case $gl_cv_func_fchmodat_works in
+      *yes) ;;
+      *) REPLACE_FCHMODAT=1;;
+    esac
   fi
 ])
diff --git a/m4/lchmod.m4 b/m4/lchmod.m4
index eeca7ac20..68dab7a4a 100644
--- a/m4/lchmod.m4
+++ b/m4/lchmod.m4
@@ -1,4 +1,4 @@
-#serial 3
+#serial 4
 
 dnl Copyright (C) 2005-2006, 2008-2020 Free Software Foundation, Inc.
 dnl This file is free software; the Free Software Foundation
@@ -6,7 +6,7 @@ dnl gives unlimited permission to copy and/or distribute it,
 dnl with or without modifications, as long as this notice is preserved.
 
 dnl From Paul Eggert.
-dnl Provide a replacement for lchmod on hosts that lack it.
+dnl Provide a replacement for lchmod on hosts that lack a working version.
 
 AC_DEFUN([gl_FUNC_LCHMOD],
 [
@@ -15,8 +15,52 @@ AC_DEFUN([gl_FUNC_LCHMOD],
   dnl Persuade glibc <sys/stat.h> to declare lchmod().
   AC_REQUIRE([AC_USE_SYSTEM_EXTENSIONS])
 
-  AC_CHECK_FUNCS_ONCE([lchmod])
-  if test $ac_cv_func_lchmod = no; then
+  AC_CHECK_FUNCS_ONCE([fchmodat lchmod lstat])
+  if test "$ac_cv_func_lchmod" = no; then
     HAVE_LCHMOD=0
+  else
+    AC_CACHE_CHECK([whether lchmod works on non-symlinks],
+      [gl_cv_func_lchmod_works],
+      [AC_RUN_IFELSE(
+         [AC_LANG_PROGRAM(
+            [
+              AC_INCLUDES_DEFAULT[
+              #ifndef S_IRUSR
+               #define S_IRUSR 0400
+              #endif
+              #ifndef S_IWUSR
+               #define S_IWUSR 0200
+              #endif
+              #ifndef S_IRWXU
+               #define S_IRWXU 0700
+              #endif
+              #ifndef S_IRWXG
+               #define S_IRWXG 0070
+              #endif
+              #ifndef S_IRWXO
+               #define S_IRWXO 0007
+              #endif
+            ]],
+            [[
+              int permissive = S_IRWXU | S_IRWXG | S_IRWXO;
+              int desired = S_IRUSR | S_IWUSR;
+              static char const f[] = "conftest.lchmod";
+              struct stat st;
+              if (creat (f, permissive) < 0)
+                return 1;
+              if (lchmod (f, desired) != 0)
+                return 1;
+              if (stat (f, &st) != 0)
+                return 1;
+              return ! ((st.st_mode & permissive) == desired);
+            ]])],
+         [gl_cv_func_lchmod_works=yes],
+         [gl_cv_func_lchmod_works=no],
+         [gl_cv_func_lchmod_works=$gl_cross_guess_normal])
+       rm -f conftest.lchmod])
+    case $gl_cv_func_lchmod_works in
+      *yes) ;;
+      *) REPLACE_LCHMOD=1;;
+    esac
   fi
 ])
diff --git a/m4/sys_stat_h.m4 b/m4/sys_stat_h.m4
index d63df9ebf..a5f35d46d 100644
--- a/m4/sys_stat_h.m4
+++ b/m4/sys_stat_h.m4
@@ -1,4 +1,4 @@
-# sys_stat_h.m4 serial 31   -*- Autoconf -*-
+# sys_stat_h.m4 serial 32   -*- Autoconf -*-
 dnl Copyright (C) 2006-2020 Free Software Foundation, Inc.
 dnl This file is free software; the Free Software Foundation
 dnl gives unlimited permission to copy and/or distribute it,
@@ -88,9 +88,11 @@ AC_DEFUN([gl_SYS_STAT_H_DEFAULTS],
   HAVE_MKNOD=1;         AC_SUBST([HAVE_MKNOD])
   HAVE_MKNODAT=1;       AC_SUBST([HAVE_MKNODAT])
   HAVE_UTIMENSAT=1;     AC_SUBST([HAVE_UTIMENSAT])
+  REPLACE_FCHMODAT=0;   AC_SUBST([REPLACE_FCHMODAT])
   REPLACE_FSTAT=0;      AC_SUBST([REPLACE_FSTAT])
   REPLACE_FSTATAT=0;    AC_SUBST([REPLACE_FSTATAT])
   REPLACE_FUTIMENS=0;   AC_SUBST([REPLACE_FUTIMENS])
+  REPLACE_LCHMOD=0;     AC_SUBST([REPLACE_LCHMOD])
   REPLACE_LSTAT=0;      AC_SUBST([REPLACE_LSTAT])
   REPLACE_MKDIR=0;      AC_SUBST([REPLACE_MKDIR])
   REPLACE_MKFIFO=0;     AC_SUBST([REPLACE_MKFIFO])
diff --git a/modules/fchmodat b/modules/fchmodat
index 7a5e1a688..b0f06e862 100644
--- a/modules/fchmodat
+++ b/modules/fchmodat
@@ -12,17 +12,21 @@ sys_stat
 extensions
 at-internal     [test $HAVE_FCHMODAT = 0]
 dosname         [test $HAVE_FCHMODAT = 0]
-errno           [test $HAVE_FCHMODAT = 0]
+errno           [test $HAVE_FCHMODAT = 0 || test $REPLACE_FCHMODAT = 1]
 extern-inline   [test $HAVE_FCHMODAT = 0]
 fchdir          [test $HAVE_FCHMODAT = 0]
-fcntl-h         [test $HAVE_FCHMODAT = 0]
+fcntl-h         [test $HAVE_FCHMODAT = 0 || test $REPLACE_FCHMODAT = 1]
+fstatat         [test $REPLACE_FCHMODAT = 1]
+intprops        [test $HAVE_FCHMODAT = 0 || test $REPLACE_FCHMODAT = 1]
+lchmod          [test $HAVE_FCHMODAT = 0]
 openat-die      [test $HAVE_FCHMODAT = 0]
 openat-h        [test $HAVE_FCHMODAT = 0]
 save-cwd        [test $HAVE_FCHMODAT = 0]
+unistd          [test $HAVE_FCHMODAT = 0 || test $REPLACE_FCHMODAT = 1]
 
 configure.ac:
 gl_FUNC_FCHMODAT
-if test $HAVE_FCHMODAT = 0; then
+if test $HAVE_FCHMODAT = 0 || test $REPLACE_FCHMODAT = 1; then
   AC_LIBOBJ([fchmodat])
 fi
 gl_MODULE_INDICATOR([fchmodat]) dnl for lib/openat.h
diff --git a/modules/lchmod b/modules/lchmod
index a0ac7a534..e1054f6a4 100644
--- a/modules/lchmod
+++ b/modules/lchmod
@@ -2,17 +2,28 @@ Description:
 lchmod that is actually chmod (!) on hosts lacking lchmod
 
 Files:
+lib/lchmod.c
 m4/lchmod.m4
 
 Depends-on:
-sys_stat
+errno         [test $HAVE_LCHMOD = 0 || test $REPLACE_LCHMOD = 1]
 extensions
+fcntl-h       [test $HAVE_LCHMOD = 0 || test $REPLACE_LCHMOD = 1]
+fchmodat      [test $HAVE_LCHMOD = 0 || test $REPLACE_LCHMOD = 1]
+intprops      [test $HAVE_LCHMOD = 0 || test $REPLACE_LCHMOD = 1]
+lstat         [test $HAVE_LCHMOD = 0 || test $REPLACE_LCHMOD = 1]
+sys_stat
+unistd        [test $HAVE_LCHMOD = 0 || test $REPLACE_LCHMOD = 1]
 
 configure.ac:
 gl_FUNC_LCHMOD
+if test $HAVE_LCHMOD = 0 || test $REPLACE_LCHMOD = 1; then
+  AC_LIBOBJ([lchmod])
+fi
 gl_SYS_STAT_MODULE_INDICATOR([lchmod])
 
 Makefile.am:
+lib_SOURCES += lchmod.c
 
 Include:
 <sys/stat.h>
diff --git a/modules/sys_stat b/modules/sys_stat
index 867cc8549..93e0cf07e 100644
--- a/modules/sys_stat
+++ b/modules/sys_stat
@@ -59,9 +59,11 @@ sys/stat.h: sys_stat.in.h $(top_builddir)/config.status $(CXXDEFS_H) $(ARG_NONNU
 	      -e 's|@''HAVE_MKNOD''@|$(HAVE_MKNOD)|g' \
 	      -e 's|@''HAVE_MKNODAT''@|$(HAVE_MKNODAT)|g' \
 	      -e 's|@''HAVE_UTIMENSAT''@|$(HAVE_UTIMENSAT)|g' \
+	      -e 's|@''REPLACE_FCHMODAT''@|$(REPLACE_FCHMODAT)|g' \
 	      -e 's|@''REPLACE_FSTAT''@|$(REPLACE_FSTAT)|g' \
 	      -e 's|@''REPLACE_FSTATAT''@|$(REPLACE_FSTATAT)|g' \
 	      -e 's|@''REPLACE_FUTIMENS''@|$(REPLACE_FUTIMENS)|g' \
+	      -e 's|@''REPLACE_LCHMOD''@|$(REPLACE_LCHMOD)|g' \
 	      -e 's|@''REPLACE_LSTAT''@|$(REPLACE_LSTAT)|g' \
 	      -e 's|@''REPLACE_MKDIR''@|$(REPLACE_MKDIR)|g' \
 	      -e 's|@''REPLACE_MKFIFO''@|$(REPLACE_MKFIFO)|g' \
diff --git a/tests/test-fchmodat.c b/tests/test-fchmodat.c
index 395b2b78e..df0f604d8 100644
--- a/tests/test-fchmodat.c
+++ b/tests/test-fchmodat.c
@@ -22,6 +22,8 @@
 SIGNATURE_CHECK (fchmodat, int, (int, const char *, mode_t, int));
 
 #include <errno.h>
+#include <fcntl.h>
+#include <sys/stat.h>
 #include <unistd.h>
 
 #include "macros.h"
@@ -42,5 +44,13 @@ main (void)
     ASSERT (errno == EBADF);
   }
 
+  /* Test that fchmodat works on non-symlinks, when given
+     the AT_SYMLINK_NOFOLLOW flag.  */
+  {
+    ASSERT (close (creat ("file", 0600)) == 0);
+    ASSERT (fchmodat (AT_FDCWD, "file", 0700, AT_SYMLINK_NOFOLLOW) == 0);
+    ASSERT (unlink ("file") == 0);
+  }
+
   return 0;
 }
-- 
2.24.1


^ permalink raw reply	[flat|nested] 21+ messages in thread

* Re: bug#39236: [musl] coreutils cp mishandles error return from lchmod
  2020-02-08  0:37           ` Paul Eggert
@ 2020-02-12 11:50             ` Florian Weimer
  2020-02-12 13:05               ` Rich Felker
  0 siblings, 1 reply; 21+ messages in thread
From: Florian Weimer @ 2020-02-12 11:50 UTC (permalink / raw)
  To: Paul Eggert; +Cc: Rich Felker, Florian Weimer, Gnulib bugs, musl, 39236

* Paul Eggert:

> On 1/22/20 2:05 PM, Rich Felker wrote:
>> I think we're approaching a consensus that glibc should fix this too,
>> so then it would just be gnulib matching the fix.
>
> I installed the attached patch to Gnulib in preparation for the upcoming 
> glibc fix. The patch causes fchmodat with AT_SYMLINK_NOFOLLOW to work on 
> non-symlinks, and similarly for lchmod on non-symlinks. The idea is to 
> avoid this sort of problem in the future, and to let Coreutils etc. work 
> on older platforms as if glibc 2.32 (or whatever) is already in place.

The lchmod implementation based on /proc tickles an XFS bug:

  <https://sourceware.org/ml/libc-alpha/2020-02/msg00467.html>

^ permalink raw reply	[flat|nested] 21+ messages in thread

* Re: bug#39236: [musl] coreutils cp mishandles error return from lchmod
  2020-02-12 11:50             ` Florian Weimer
@ 2020-02-12 13:05               ` Rich Felker
  2020-02-12 19:07                 ` Rich Felker
  0 siblings, 1 reply; 21+ messages in thread
From: Rich Felker @ 2020-02-12 13:05 UTC (permalink / raw)
  To: Florian Weimer; +Cc: Paul Eggert, Florian Weimer, Gnulib bugs, musl, 39236

On Wed, Feb 12, 2020 at 12:50:19PM +0100, Florian Weimer wrote:
> * Paul Eggert:
> 
> > On 1/22/20 2:05 PM, Rich Felker wrote:
> >> I think we're approaching a consensus that glibc should fix this too,
> >> so then it would just be gnulib matching the fix.
> >
> > I installed the attached patch to Gnulib in preparation for the upcoming 
> > glibc fix. The patch causes fchmodat with AT_SYMLINK_NOFOLLOW to work on 
> > non-symlinks, and similarly for lchmod on non-symlinks. The idea is to 
> > avoid this sort of problem in the future, and to let Coreutils etc. work 
> > on older platforms as if glibc 2.32 (or whatever) is already in place.
> 
> The lchmod implementation based on /proc tickles an XFS bug:
> 
>   <https://sourceware.org/ml/libc-alpha/2020-02/msg00467.html>

Uhg, why does Linux even let the fs driver see whether the chmod is
being performed via a filename, O_PATH fd, or magic symlink in /proc?
It should just be an operation on the inode.

Rich

^ permalink raw reply	[flat|nested] 21+ messages in thread

* Re: bug#39236: [musl] coreutils cp mishandles error return from lchmod
  2020-02-12 13:05               ` Rich Felker
@ 2020-02-12 19:07                 ` Rich Felker
  2020-02-12 19:13                   ` Florian Weimer
  2020-02-12 19:59                   ` A. Wilcox
  0 siblings, 2 replies; 21+ messages in thread
From: Rich Felker @ 2020-02-12 19:07 UTC (permalink / raw)
  To: Florian Weimer; +Cc: Paul Eggert, Florian Weimer, Gnulib bugs, musl, 39236

On Wed, Feb 12, 2020 at 08:05:55AM -0500, Rich Felker wrote:
> On Wed, Feb 12, 2020 at 12:50:19PM +0100, Florian Weimer wrote:
> > * Paul Eggert:
> > 
> > > On 1/22/20 2:05 PM, Rich Felker wrote:
> > >> I think we're approaching a consensus that glibc should fix this too,
> > >> so then it would just be gnulib matching the fix.
> > >
> > > I installed the attached patch to Gnulib in preparation for the upcoming 
> > > glibc fix. The patch causes fchmodat with AT_SYMLINK_NOFOLLOW to work on 
> > > non-symlinks, and similarly for lchmod on non-symlinks. The idea is to 
> > > avoid this sort of problem in the future, and to let Coreutils etc. work 
> > > on older platforms as if glibc 2.32 (or whatever) is already in place.
> > 
> > The lchmod implementation based on /proc tickles an XFS bug:
> > 
> >   <https://sourceware.org/ml/libc-alpha/2020-02/msg00467.html>
> 
> Uhg, why does Linux even let the fs driver see whether the chmod is
> being performed via a filename, O_PATH fd, or magic symlink in /proc?
> It should just be an operation on the inode.

OK, I don't think it's actually clear from the test that the use of
the magic symlink is the cause. It's plausible that XFS just always
returns failure on success for this operation, and I don't have XFS to
test with.

Note that in any case, musl's lchmod/fchmodat is not affected since it
always refuses to change symlink modes; I did this because I was
worried that chmod on the magic symlink in /proc might pass through
not just to the symlink it refers to, but to the symlink target if one
exists. With current kernel versions it seems that does not happen; is
it safe to assume it doesn't?

Further, I've found some inconsistent behavior with ext4: chmod on the
magic symlink fails with EOPNOTSUPP as in Florian's test, but fchmod
on the O_PATH fd succeeds and changes the symlink mode. This is with
5.4. Cany anyone else confirm this? Is it a problem?

Rich

^ permalink raw reply	[flat|nested] 21+ messages in thread

* Re: bug#39236: [musl] coreutils cp mishandles error return from lchmod
  2020-02-12 19:07                 ` Rich Felker
@ 2020-02-12 19:13                   ` Florian Weimer
  2020-02-12 19:59                   ` A. Wilcox
  1 sibling, 0 replies; 21+ messages in thread
From: Florian Weimer @ 2020-02-12 19:13 UTC (permalink / raw)
  To: Rich Felker; +Cc: Paul Eggert, Gnulib bugs, musl, 39236

* Rich Felker:

> Note that in any case, musl's lchmod/fchmodat is not affected since it
> always refuses to change symlink modes; I did this because I was
> worried that chmod on the magic symlink in /proc might pass through
> not just to the symlink it refers to, but to the symlink target if one
> exists. With current kernel versions it seems that does not happen; is
> it safe to assume it doesn't?

I saw it happen with sshfs over FUSE. 8-/

Yet another reason to put in a check before performing the chmod.

> Further, I've found some inconsistent behavior with ext4: chmod on the
> magic symlink fails with EOPNOTSUPP as in Florian's test, but fchmod
> on the O_PATH fd succeeds and changes the symlink mode. This is with
> 5.4. Cany anyone else confirm this? Is it a problem?

Interesting. Let me update the other thread.

^ permalink raw reply	[flat|nested] 21+ messages in thread

* Re: bug#39236: [musl] coreutils cp mishandles error return from lchmod
  2020-02-12 19:07                 ` Rich Felker
  2020-02-12 19:13                   ` Florian Weimer
@ 2020-02-12 19:59                   ` A. Wilcox
  2020-02-12 20:56                     ` Rich Felker
  1 sibling, 1 reply; 21+ messages in thread
From: A. Wilcox @ 2020-02-12 19:59 UTC (permalink / raw)
  To: musl

[-- Attachment #1.1: Type: text/plain, Size: 1793 bytes --]

On 12/02/2020 13:07, Rich Felker wrote:
> On Wed, Feb 12, 2020 at 08:05:55AM -0500, Rich Felker wrote:
>> On Wed, Feb 12, 2020 at 12:50:19PM +0100, Florian Weimer wrote:
>>> * Paul Eggert:
>>>
>>>> On 1/22/20 2:05 PM, Rich Felker wrote:
>>>>> I think we're approaching a consensus that glibc should fix this too,
>>>>> so then it would just be gnulib matching the fix.
>>>>
>>>> I installed the attached patch to Gnulib in preparation for the upcoming 
>>>> glibc fix. The patch causes fchmodat with AT_SYMLINK_NOFOLLOW to work on 
>>>> non-symlinks, and similarly for lchmod on non-symlinks. The idea is to 
>>>> avoid this sort of problem in the future, and to let Coreutils etc. work 
>>>> on older platforms as if glibc 2.32 (or whatever) is already in place.
>>>
>>> The lchmod implementation based on /proc tickles an XFS bug:
>>>
>>>   <https://sourceware.org/ml/libc-alpha/2020-02/msg00467.html>
>>
>> Uhg, why does Linux even let the fs driver see whether the chmod is
>> being performed via a filename, O_PATH fd, or magic symlink in /proc?
>> It should just be an operation on the inode.
> 
> OK, I don't think it's actually clear from the test that the use of
> the magic symlink is the cause. It's plausible that XFS just always
> returns failure on success for this operation, and I don't have XFS to
> test with.


My root fs is XFS, but I only have musl to test with.  Is there a test
case I can run on musl to determine the behaviour of XFS for you?

The only glibc distribution that supports my platform is Void, so I
don't know if the Void glibc spin in a chroot would be sufficient if
there is no way to do this from a musl system.

Best,
--arw


-- 
A. Wilcox (awilfox)
Project Lead, Adélie Linux
https://www.adelielinux.org


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

^ permalink raw reply	[flat|nested] 21+ messages in thread

* Re: bug#39236: [musl] coreutils cp mishandles error return from lchmod
  2020-02-12 19:59                   ` A. Wilcox
@ 2020-02-12 20:56                     ` Rich Felker
  0 siblings, 0 replies; 21+ messages in thread
From: Rich Felker @ 2020-02-12 20:56 UTC (permalink / raw)
  To: A. Wilcox; +Cc: musl

On Wed, Feb 12, 2020 at 01:59:09PM -0600, A. Wilcox wrote:
> On 12/02/2020 13:07, Rich Felker wrote:
> > On Wed, Feb 12, 2020 at 08:05:55AM -0500, Rich Felker wrote:
> >> On Wed, Feb 12, 2020 at 12:50:19PM +0100, Florian Weimer wrote:
> >>> * Paul Eggert:
> >>>
> >>>> On 1/22/20 2:05 PM, Rich Felker wrote:
> >>>>> I think we're approaching a consensus that glibc should fix this too,
> >>>>> so then it would just be gnulib matching the fix.
> >>>>
> >>>> I installed the attached patch to Gnulib in preparation for the upcoming 
> >>>> glibc fix. The patch causes fchmodat with AT_SYMLINK_NOFOLLOW to work on 
> >>>> non-symlinks, and similarly for lchmod on non-symlinks. The idea is to 
> >>>> avoid this sort of problem in the future, and to let Coreutils etc. work 
> >>>> on older platforms as if glibc 2.32 (or whatever) is already in place.
> >>>
> >>> The lchmod implementation based on /proc tickles an XFS bug:
> >>>
> >>>   <https://sourceware.org/ml/libc-alpha/2020-02/msg00467.html>
> >>
> >> Uhg, why does Linux even let the fs driver see whether the chmod is
> >> being performed via a filename, O_PATH fd, or magic symlink in /proc?
> >> It should just be an operation on the inode.
> > 
> > OK, I don't think it's actually clear from the test that the use of
> > the magic symlink is the cause. It's plausible that XFS just always
> > returns failure on success for this operation, and I don't have XFS to
> > test with.
> 
> 
> My root fs is XFS, but I only have musl to test with.  Is there a test
> case I can run on musl to determine the behaviour of XFS for you?
> 
> The only glibc distribution that supports my platform is Void, so I
> don't know if the Void glibc spin in a chroot would be sufficient if
> there is no way to do this from a musl system.

We've since found (in the other thread musl list isn't CC'd on) that
this does not seem to be XFS-specific but happens in ext4 too, and is
probably a high level vfs bug.

Rich

^ permalink raw reply	[flat|nested] 21+ messages in thread

end of thread, back to index

Thread overview: 21+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2020-01-22 14:15 [musl] coreutils cp mishandles error return from lchmod Rich Felker
2020-01-22 14:34 ` Florian Weimer
2020-01-22 14:42   ` Rich Felker
2020-01-22 15:08     ` Florian Weimer
2020-01-22 15:15       ` Rich Felker
2020-01-22 15:32         ` Florian Weimer
2020-01-22 16:07           ` Rich Felker
2020-01-22 16:19             ` Florian Weimer
2020-01-22 17:15               ` Rich Felker
2020-01-22 20:48                 ` Florian Weimer
2020-01-22 20:56                   ` Rich Felker
2020-01-22 21:05                     ` Florian Weimer
2020-01-22 21:55       ` bug#39236: " Paul Eggert
2020-01-22 22:05         ` Rich Felker
2020-02-08  0:37           ` Paul Eggert
2020-02-12 11:50             ` Florian Weimer
2020-02-12 13:05               ` Rich Felker
2020-02-12 19:07                 ` Rich Felker
2020-02-12 19:13                   ` Florian Weimer
2020-02-12 19:59                   ` A. Wilcox
2020-02-12 20:56                     ` Rich Felker

mailing list of musl libc

Archives are clonable: git clone --mirror http://inbox.vuxu.org/musl

Example config snippet for mirrors

Newsgroup available over NNTP:
	nntp://inbox.vuxu.org/vuxu.archive.musl


AGPL code for this site: git clone https://public-inbox.org/public-inbox.git