Bug#368645: CVE-2006-2313, CVE-2006-2314: encoding conflicts - sarge security update finished

2006-05-29 Thread Martin Pitt
Hi Joey,

Martin Schulze [2006-05-28 19:37 +0200]:
  [1] http://people.debian.org/~mpitt/psql-sarge/
  [2] 
  http://people.debian.org/~mpitt/psql-sarge/postgresql_7.4.7-6sarge2.debdiff
 
 Thanks a lot.  However, could you redo the (source) package without
 the arch crap inside?

There is no arch stuff inside (I don't even use arch any more). I also
cleaned the debdiff (I just checked again).
However, the -sarge1 version had arch stuff, maybe you did a debdiff
on your own and stumbled over that?

Martin

-- 
Martin Pitthttp://www.piware.de
Ubuntu Developer   http://www.ubuntu.com
Debian Developer   http://www.debian.org

In a world without walls and fences, who needs Windows and Gates?


signature.asc
Description: Digital signature


Bug#368645: CVE-2006-2313, CVE-2006-2314: encoding conflicts - sarge security update finished

2006-05-29 Thread Martin Schulze
Martin Pitt wrote:
 Hi Joey,
 
 Martin Schulze [2006-05-28 19:37 +0200]:
   [1] http://people.debian.org/~mpitt/psql-sarge/
   [2] 
   http://people.debian.org/~mpitt/psql-sarge/postgresql_7.4.7-6sarge2.debdiff
  
  Thanks a lot.  However, could you redo the (source) package without
  the arch crap inside?
 
 There is no arch stuff inside (I don't even use arch any more). I also
 cleaned the debdiff (I just checked again).
 However, the -sarge1 version had arch stuff, maybe you did a debdiff
 on your own and stumbled over that?

Yup.  I see.  In that case the arch stuff should be kept so the patch
is not cluttered.

Regards,

Joey

-- 
Ten years and still binary compatible.  -- XFree86

Please always Cc to me when replying to me on the lists.


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]



Bug#368645: CVE-2006-2313, CVE-2006-2314: encoding conflicts - sarge security update finished

2006-05-29 Thread Martin Pitt
Hi,

Martin Schulze [2006-05-29 15:25 +0200]:
 Martin Pitt wrote:
  Hi Joey,
  
  Martin Schulze [2006-05-28 19:37 +0200]:
[1] http://people.debian.org/~mpitt/psql-sarge/
[2] 
http://people.debian.org/~mpitt/psql-sarge/postgresql_7.4.7-6sarge2.debdiff
   
   Thanks a lot.  However, could you redo the (source) package without
   the arch crap inside?
  
  There is no arch stuff inside (I don't even use arch any more). I also
  cleaned the debdiff (I just checked again).
  However, the -sarge1 version had arch stuff, maybe you did a debdiff
  on your own and stumbled over that?
 
 Yup.  I see.  In that case the arch stuff should be kept so the patch
 is not cluttered.

I can't, sorry. I killed the arch repo months ago. The debdiff in [2]
does not contain arch spewage.

Martin

-- 
Martin Pitthttp://www.piware.de
Ubuntu Developer   http://www.ubuntu.com
Debian Developer   http://www.debian.org

In a world without walls and fences, who needs Windows and Gates?


signature.asc
Description: Digital signature


Bug#368645: CVE-2006-2313, CVE-2006-2314: encoding conflicts - sarge security update finished

2006-05-28 Thread Martin Pitt
Hi security team,

I backported the relevant changes from 7.4.13 and put the sarge
security update to [1]. This time, just putting 7.4.13 into
sarge-security would even have been safer IMHO, and that's what users
would want anyway, but we already had this discussion several times,
so I only ported the security fixes and a very simple, but important
bug fix.

The debdiff is available [2], but believe me, you do not really want
to look at it. You have been warned! :)

The package passes the upstream test suite, the same patches thrown
onto 7.4.8 (which Ubuntu uses in version 5.04) pass my own test suite
in postgresql-common, and the exploit does not work any more, so I'm
fairly sure that it doesn't break too much.

Please feel free to just upload the provided package, or tell me how
to proceed.

Thank you!

Martin

[1] http://people.debian.org/~mpitt/psql-sarge/
[2] http://people.debian.org/~mpitt/psql-sarge/postgresql_7.4.7-6sarge2.debdiff

-- 
Martin Pitthttp://www.piware.de
Ubuntu Developer   http://www.ubuntu.com
Debian Developer   http://www.debian.org

In a world without walls and fences, who needs Windows and Gates?


signature.asc
Description: Digital signature


Bug#368645: CVE-2006-2313, CVE-2006-2314: encoding conflicts - sarge security update finished

2006-05-28 Thread Martin Schulze
Martin Pitt wrote:
 Hi security team,
 
 I backported the relevant changes from 7.4.13 and put the sarge
 security update to [1]. This time, just putting 7.4.13 into
 sarge-security would even have been safer IMHO, and that's what users
 would want anyway, but we already had this discussion several times,
 so I only ported the security fixes and a very simple, but important
 bug fix.
 
 The debdiff is available [2], but believe me, you do not really want
 to look at it. You have been warned! :)
 
 The package passes the upstream test suite, the same patches thrown
 onto 7.4.8 (which Ubuntu uses in version 5.04) pass my own test suite
 in postgresql-common, and the exploit does not work any more, so I'm
 fairly sure that it doesn't break too much.
 
 Please feel free to just upload the provided package, or tell me how
 to proceed.
 
 Thank you!
 
 Martin
 
 [1] http://people.debian.org/~mpitt/psql-sarge/
 [2] 
 http://people.debian.org/~mpitt/psql-sarge/postgresql_7.4.7-6sarge2.debdiff

Thanks a lot.  However, could you redo the (source) package without
the arch crap inside?

Regards,

Joey

-- 
A mathematician is a machine for converting coffee into theorems.   Paul Erdös

Please always Cc to me when replying to me on the lists.



Bug#368645: CVE-2006-2313, CVE-2006-2314: encoding conflicts

2006-05-25 Thread Martin Schulze
Martin Pitt wrote:
 Hi Florian, hi security team, hi everyone else,
 
 just for the record, sid has updated packages already.
 
 I'm 70% into completing the security update for sarge. However, due to
 the nature of the vulns, the patches are enormous, and thus require
 meticulous porting and testing.
 
 Unfortunately I will be away from now until Sunday. I hope to have
 fixed packages ready on Sunday. I will report back when I'm done.

Oh dear!  Thanks a lot.

Regards,

Joey

Given enough thrust pigs will fly, but it's not necessarily a good idea.

Please always Cc to me when replying to me on the lists.


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]



Bug#368645: CVE-2006-2313, CVE-2006-2314: encoding conflicts

2006-05-24 Thread Martin Pitt
Hi Florian, hi security team, hi everyone else,

just for the record, sid has updated packages already.

I'm 70% into completing the security update for sarge. However, due to
the nature of the vulns, the patches are enormous, and thus require
meticulous porting and testing.

Unfortunately I will be away from now until Sunday. I hope to have
fixed packages ready on Sunday. I will report back when I'm done.

Thanks,

Martin

-- 
Martin Pitthttp://www.piware.de
Ubuntu Developer   http://www.ubuntu.com
Debian Developer   http://www.debian.org

In a world without walls and fences, who needs Windows and Gates?


signature.asc
Description: Digital signature


Bug#368645: CVE-2006-2313, CVE-2006-2314: encoding conflicts

2006-05-23 Thread Florian Weimer
Package: postgresql
Version: 7.4.7-6sarge1
Tags: security
Severity: grave

A couple of PostgreSQL issues have been disclosed today:

  http://www.postgresql.org/docs/techdocs.52

My analysis so far:

* CVE-2006-2313

High impact (because UTF-8 is affected and widely used).  Fix is
straightforward as far as UTF-8 is concerned, but will break some
applications which write certain forms of invalid UTF-8 to the
database.  If necessary, a dump and reload to switch to SQL_ASCII on
the server side will fix this.  However, PostgreSQL already rejects
some forms of invalid UTF-8.  Therefore, a change

I don't know the impact on other multibyte encodings; it's probably
necessary to ask upstream.

* CVE-2006-2314

This is the really interesting one.  It's restricted to certain
multi-byte encodings (that's why I think this bug is less severe, all
things considered).  No real fix is possible as long as we preserve
the interface.  The upstream fix outlawing \' breaks tons of legacy
PHP applications, but I have no better idea how to address it. 8-(

On the libpq side, I'd use static __thread instead of static for
the globals.  That way, we gain at least some thread safety.

(Unless someone objects, I'm going to clone this for the various
PostgreSQL packages.)


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]



Bug#368645: CVE-2006-2313, CVE-2006-2314: encoding conflicts

2006-05-23 Thread Peter Eisentraut
Florian Weimer wrote:
 (Unless someone objects, I'm going to clone this for the various
 PostgreSQL packages.)

Packages are already being uploaded, so don't waste everyone's time.


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]



Bug#368645: CVE-2006-2313, CVE-2006-2314: encoding conflicts

2006-05-23 Thread Peter Eisentraut
Peter Eisentraut wrote:
 Florian Weimer wrote:
  (Unless someone objects, I'm going to clone this for the various
  PostgreSQL packages.)

 Packages are already being uploaded, so don't waste everyone's time.

Correction: packages have already been uploaded, so we only need to wait 
for the security team's approval of the stable upload.

(Yes, there is a secret club that coordinates these things before the 
publication of the security issue.)


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]