On Thu, Nov 18, 2021 at 8:38 AM Paul Eggert wrote:
> I spotted a SELinux security-context race introduced by the circa-2012
> fix for Bug#11100, and installed the attached patch into coreutils
> master. This also gets rid of a label and goto (which is what led me to
> find the issue).
Nice!
I spotted a SELinux security-context race introduced by the circa-2012
fix for Bug#11100, and installed the attached patch into coreutils
master. This also gets rid of a label and goto (which is what led me to
find the issue).From d0f035fc64fb348cb092fbb6ae7e8ce76b4d82db Mon Sep 17 00:00:00
* Jim Meyering (j...@meyering.net) [20120504 17:30]:
If there's a bugzilla reference for this, let me know
and I'll add it to the commit log.
There is, but as it's a SLES bug it's only open for SUSE employees and
customers and thus useless for a coreutils commit log. I'll instead
reference the
Philipp Thomas wrote:
* Jim Meyering (j...@meyering.net) [20120504 17:30]:
If there's a bugzilla reference for this, let me know
and I'll add it to the commit log.
There is, but as it's a SLES bug it's only open for SUSE employees and
customers and thus useless for a coreutils commit log.
Jim Meyering wrote:
Philipp Thomas wrote:
* Jim Meyering (j...@meyering.net) [20120328 18:09]:
At first glance, that might be reasonable: the additional open
is incurred only after a failed stat.
I'll look more closely in a week or two if no one else investigates.
Ping, it's been more than
Pádraig Brady wrote:
I can't think of any issue with this.
Code looks good.
Test triggers the new condition.
Thanks for the review.
I've squashed the test-adding commit onto the fix, added this sentence
to NEWS:
With NFS attribute caching, the condition
was particularly easy to
* Jim Meyering (j...@meyering.net) [20120328 18:09]:
At first glance, that might be reasonable: the additional open
is incurred only after a failed stat.
I'll look more closely in a week or two if no one else investigates.
Ping, it's been more than two weeks and I'm being bugged for a
Philipp Thomas wrote:
* Jim Meyering (j...@meyering.net) [20120328 18:09]:
At first glance, that might be reasonable: the additional open
is incurred only after a failed stat.
I'll look more closely in a week or two if no one else investigates.
Ping, it's been more than two weeks and I'm
On 05/04/2012 08:47 AM, Jim Meyering wrote:
* src/copy.c (copy_reg): In a narrow race (stat sees dest, yet
open-without-O_CREAT fails with ENOENT), retry the open with O_CREAT.
Reported by Philipp Thomas and Neil F. Brown in
http://debbugs.gnu.org/11100
Question - when we retry with adding
Eric Blake wrote:
On 05/04/2012 08:47 AM, Jim Meyering wrote:
* src/copy.c (copy_reg): In a narrow race (stat sees dest, yet
open-without-O_CREAT fails with ENOENT), retry the open with O_CREAT.
Reported by Philipp Thomas and Neil F. Brown in
http://debbugs.gnu.org/11100
Question - when
Eric Blake wrote:
On 05/04/2012 08:47 AM, Jim Meyering wrote:
* src/copy.c (copy_reg): In a narrow race (stat sees dest, yet
open-without-O_CREAT fails with ENOENT), retry the open with O_CREAT.
Reported by Philipp Thomas and Neil F. Brown in
http://debbugs.gnu.org/11100
Question - when we
NeilBrown wrote:
On Tue, 27 Mar 2012 15:40:25 +0200 Jim Meyering j...@meyering.net wrote:
Philipp Thomas wrote:
I'd like to pass on observations from my collegue Neil Brown:
in src/copy.c, copy_reg() is passed bool *new_dst.
This is 'false' if the file already exists, in which case it
On Wed, 28 Mar 2012 18:07:51 +0200 Jim Meyering j...@meyering.net wrote:
NeilBrown wrote:
On Tue, 27 Mar 2012 15:40:25 +0200 Jim Meyering j...@meyering.net wrote:
Philipp Thomas wrote:
I'd like to pass on observations from my collegue Neil Brown:
in src/copy.c, copy_reg() is
On 03/28/2012 09:07 AM, Jim Meyering wrote:
if ((use_stat
- ? stat (dst_name, dst_sb)
+ ? (stat (dst_name, dst_sb) 0 ? -1 :
+(fd = open (dst_name, O_WRONLY)) 0 ? -1 : 0)
: lstat (dst_name, dst_sb))
!= 0)
At first glance, that
I'd like to pass on observations from my collegue Neil Brown:
in src/copy.c, copy_reg() is passed bool *new_dst.
This is 'false' if the file already exists, in which case it attempts to
open the file with O_WRONLY | O_TRUNC | O_BINARY.
If it is 'true', only then does it use O_CREAT (and others).
Philipp Thomas wrote:
I'd like to pass on observations from my collegue Neil Brown:
in src/copy.c, copy_reg() is passed bool *new_dst.
This is 'false' if the file already exists, in which case it attempts to
open the file with O_WRONLY | O_TRUNC | O_BINARY.
If it is 'true', only then does
Hi Jim,
* Jim Meyering (j...@meyering.net) [20120327 15:40]:
If you can find a way to make cp work sensibly in your specific case,
yet without impacting any other use case, please let us know.
Thanks for the clarification! In that light though I doubt there is a way :(
Philipp
On 03/27/2012 01:58 PM, Philipp Thomas wrote:
I'd like to pass on observations from my collegue Neil Brown:
in src/copy.c, copy_reg() is passed bool *new_dst.
This is 'false' if the file already exists, in which case it attempts to
open the file with O_WRONLY | O_TRUNC | O_BINARY.
If it
On 03/27/2012 05:58 AM, Philipp Thomas wrote:
The above mentioned code assumes that the file still
exists. This is racy - particularly for NFS where deletions from other
clients can take a while to appear.
*NEW_DST is a bit more complicated than that. At least for part
of the code it means
On Tue, 27 Mar 2012 15:40:25 +0200 Jim Meyering j...@meyering.net wrote:
Philipp Thomas wrote:
I'd like to pass on observations from my collegue Neil Brown:
in src/copy.c, copy_reg() is passed bool *new_dst.
This is 'false' if the file already exists, in which case it attempts to
20 matches
Mail list logo
