All of lore.kernel.org
 help / color / mirror / Atom feed
From: Eric Blake <eblake@redhat.com>
To: Peter Maydell <peter.maydell@linaro.org>
Cc: QEMU Trivial <qemu-trivial@nongnu.org>,
	Paolo Bonzini <pbonzini@redhat.com>,
	QEMU Developers <qemu-devel@nongnu.org>,
	Don Slutz <dslutz@verizon.com>,
	Patch Tracking <patches@linaro.org>
Subject: Re: [Qemu-trivial] [Qemu-devel] [PATCH] configure: Put tempfiles in subdir so we can clean up libtool files
Date: Tue, 06 May 2014 09:43:57 -0600	[thread overview]
Message-ID: <5369033D.5060505@redhat.com> (raw)
In-Reply-To: <CAFEAcA_muHezDUmdcvkEqp0pFmCBJ55kUXrSZkuUv05vxpJybw@mail.gmail.com>

[-- Attachment #1: Type: text/plain, Size: 1836 bytes --]

On 05/06/2014 08:53 AM, Peter Maydell wrote:

>> # Create a (secure) tmp directory for tmp files.
>>
>> {
>>   tmp=`(umask 077 && mktemp -d "./confXXXXXX") 2>/dev/null` &&
>>   test -d "$tmp"
>> }  ||
>> {
>>   tmp=./conf$$-$RANDOM
>>   (umask 077 && mkdir "$tmp")
>> } || as_fn_error $? "cannot create a temporary directory in ." "$LINENO" 5
>> ac_tmp=$tmp
> 
> Yuck.
> 
>> The use of $$ and $RANDOM is safe (even on shells that lack $RANDOM)
>> because of the fact that mkdir is atomic and the umask is correctly set
>> prior to the mkdir.
> 
> I dislike the use of $RANDOM, because it means we behave
> inconsistently. If it's OK for $RANDOM to expand to "" then we
> should just not use it at all, because that's OK and the same
> everywhere.

It's okay for $RANDOM to expand to "" in the fallback code, for the
platforms that lack mktemp(1); most developers are on a platform that
have mktemp.  The use of $RANDOM makes it harder for an attacker to
pre-create a competing file by the same name, but does not add any
security; so omitting $RANDOM for the fallback path doesn't hurt if you
are that bothered by seeing it present in a dash script.

> 
> Similarly, if it's OK not to use mktemp on some systems,
> we should use the same non-mktemp code everywhere.

The fallback is not ideal, but tolerable.  It's still better to try and
use mktemp where it exists.

> 
> We could sidestep this rubbish by not trying to put our temp
> files in /tmp/, and instead just put them in the build directory
> (ie ./conf-temps/ or something similar, which we blow away
> and recreate every time).

Yes, using a different location for temporary files and avoiding /tmp
might also work.

-- 
Eric Blake   eblake redhat com    +1-919-301-3266
Libvirt virtualization library http://libvirt.org


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 604 bytes --]

WARNING: multiple messages have this Message-ID (diff)
From: Eric Blake <eblake@redhat.com>
To: Peter Maydell <peter.maydell@linaro.org>
Cc: QEMU Trivial <qemu-trivial@nongnu.org>,
	Paolo Bonzini <pbonzini@redhat.com>,
	QEMU Developers <qemu-devel@nongnu.org>,
	Don Slutz <dslutz@verizon.com>,
	Patch Tracking <patches@linaro.org>
Subject: Re: [Qemu-devel] [PATCH] configure: Put tempfiles in subdir so we can clean up libtool files
Date: Tue, 06 May 2014 09:43:57 -0600	[thread overview]
Message-ID: <5369033D.5060505@redhat.com> (raw)
In-Reply-To: <CAFEAcA_muHezDUmdcvkEqp0pFmCBJ55kUXrSZkuUv05vxpJybw@mail.gmail.com>

[-- Attachment #1: Type: text/plain, Size: 1836 bytes --]

On 05/06/2014 08:53 AM, Peter Maydell wrote:

>> # Create a (secure) tmp directory for tmp files.
>>
>> {
>>   tmp=`(umask 077 && mktemp -d "./confXXXXXX") 2>/dev/null` &&
>>   test -d "$tmp"
>> }  ||
>> {
>>   tmp=./conf$$-$RANDOM
>>   (umask 077 && mkdir "$tmp")
>> } || as_fn_error $? "cannot create a temporary directory in ." "$LINENO" 5
>> ac_tmp=$tmp
> 
> Yuck.
> 
>> The use of $$ and $RANDOM is safe (even on shells that lack $RANDOM)
>> because of the fact that mkdir is atomic and the umask is correctly set
>> prior to the mkdir.
> 
> I dislike the use of $RANDOM, because it means we behave
> inconsistently. If it's OK for $RANDOM to expand to "" then we
> should just not use it at all, because that's OK and the same
> everywhere.

It's okay for $RANDOM to expand to "" in the fallback code, for the
platforms that lack mktemp(1); most developers are on a platform that
have mktemp.  The use of $RANDOM makes it harder for an attacker to
pre-create a competing file by the same name, but does not add any
security; so omitting $RANDOM for the fallback path doesn't hurt if you
are that bothered by seeing it present in a dash script.

> 
> Similarly, if it's OK not to use mktemp on some systems,
> we should use the same non-mktemp code everywhere.

The fallback is not ideal, but tolerable.  It's still better to try and
use mktemp where it exists.

> 
> We could sidestep this rubbish by not trying to put our temp
> files in /tmp/, and instead just put them in the build directory
> (ie ./conf-temps/ or something similar, which we blow away
> and recreate every time).

Yes, using a different location for temporary files and avoiding /tmp
might also work.

-- 
Eric Blake   eblake redhat com    +1-919-301-3266
Libvirt virtualization library http://libvirt.org


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 604 bytes --]

  reply	other threads:[~2014-05-06 15:44 UTC|newest]

Thread overview: 9+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2014-05-06 13:17 [Qemu-trivial] [PATCH] configure: Put tempfiles in subdir so we can clean up libtool files Peter Maydell
2014-05-06 13:17 ` [Qemu-devel] " Peter Maydell
2014-05-06 14:36 ` [Qemu-trivial] " Eric Blake
2014-05-06 14:36   ` Eric Blake
2014-05-06 14:53   ` [Qemu-trivial] " Peter Maydell
2014-05-06 14:53     ` Peter Maydell
2014-05-06 15:43     ` Eric Blake [this message]
2014-05-06 15:43       ` Eric Blake
2014-05-29  2:33   ` [Qemu-trivial] " Ed Maste

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=5369033D.5060505@redhat.com \
    --to=eblake@redhat.com \
    --cc=dslutz@verizon.com \
    --cc=patches@linaro.org \
    --cc=pbonzini@redhat.com \
    --cc=peter.maydell@linaro.org \
    --cc=qemu-devel@nongnu.org \
    --cc=qemu-trivial@nongnu.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.