> "Alfredo" == Alfredo Braunstein <[EMAIL PROTECTED]> writes:
Alfredo> Done. Should I update also some of status/ANNOUNCE or would
Alfredo> you do it (my english sux big time)?
I'll do it.
JMarc
Jean-Marc Lasgouttes wrote:
> Alfredo> I don't know what's better. I'm inclined to apply only the
> Alfredo> first patch I've sent (solving only the external inset
> Alfredo> problem - which is the most urgent) because it's less likely
> Alfredo> it will brake anything else.
>
> Yes, I suggest th
> "Alfredo" == Alfredo Braunstein <[EMAIL PROTECTED]> writes:
Alfredo> Of course, this can brake a lot of manually-defined
Alfredo> converters (using > and < for instance).
Note that we have a general_command_wrapper.py script for these kind
of things. I do not know how useful it is here, tho
Angus Leeming wrote:
>> I don't know if playing with the file name to be converted you can
>> also execute arbitrary code. Tried a bit and didn't work... but it
>> would be nicer/safer to use execvp anyways?
>
> Agreed. There is no need to retain systemcall at all. Kill it.
This would be it, exc
On Fri, Feb 06, 2004 at 08:42:47AM +0100, Alfredo Braunstein wrote:
> Andre Poenitz wrote:
>
> >> $ cat redirect.sh
> >> #! /bin/sh
> >> exec $1 > $2
> >>
> >>
> >> \converter tiff eps "sh $$s/redirect.sh('tiff2ps $$i' $$o)" ""
> >
> > That's as insecure as the original solution, isn't it?
>
>
Andre Poenitz wrote:
>> $ cat redirect.sh
>> #! /bin/sh
>> exec $1 > $2
>>
>>
>> \converter tiff eps "sh $$s/redirect.sh('tiff2ps $$i' $$o)" ""
>
> That's as insecure as the original solution, isn't it?
It doesn't seem so...
for instance a && in $1 or $2 won't be interpreted if I'm not mistak
On Thu, Feb 05, 2004 at 06:06:28PM +, Angus Leeming wrote:
> Alfredo Braunstein wrote:
> > There's one default converter that won't work:
> >
> > \converter tiff eps "tiff2ps $$i > $$o" ""
> >
> > Should we provide a script or do we just nuke it?
> >
> > Alfredo
>
> Would this work and, if
Angus Leeming wrote:
> Alfredo Braunstein wrote:
>
>> Angus Leeming wrote:
>>
>>> Would this work and, if so, is it safe?
>>
>> Seems to be ok, but I'm no expert...
>
> Ask on comp.lang.unix.whatever then...
Answer: "whatever."
They always answer like that in that ng.
Alfredo
Alfredo Braunstein wrote:
> Angus Leeming wrote:
>
>> Would this work and, if so, is it safe?
>
> Seems to be ok, but I'm no expert...
Ask on comp.lang.unix.whatever then...
>
>> $ cat redirect.sh
>> #! /bin/sh
>> exec $1 > $2
>>
>>
>> \converter tiff eps "sh $$s/redirect.sh('tiff2ps $$i'
Angus Leeming wrote:
> Would this work and, if so, is it safe?
Seems to be ok, but I'm no expert...
> $ cat redirect.sh
> #! /bin/sh
> exec $1 > $2
>
>
> \converter tiff eps "sh $$s/redirect.sh('tiff2ps $$i' $$o)" ""
Alfredo
Alfredo Braunstein wrote:
> There's one default converter that won't work:
>
> \converter tiff eps "tiff2ps $$i > $$o" ""
>
> Should we provide a script or do we just nuke it?
>
> Alfredo
Would this work and, if so, is it safe?
$ cat redirect.sh
#! /bin/sh
exec $1 > $2
\converter tiff eps "s
Angus Leeming wrote:
>> Should we do something equivalent for converters?
>
> Don't they use forkedcall already?
Nope:
[EMAIL PROTECTED] src]$ grep -l Systemcall *.C */*.C */*/*.C
Chktex.C
converter.C
LaTeX.C
lyx_cb.C
vc-backend.C
insets/insetgraphics.C
support/filetools.C
support/systemcall.C
Alfredo Braunstein wrote:
> Can someone test the current patch with documents with external
> insets to see if all works OK?
>
> Attached also a testcase showing the bad behaviour before the patch.
Can't happen with 1.4.x AFAIK.
> Should we do something equivalent for converters?
Don't they us
Lars Gullik BjÃnnes wrote:
> Alfredo Braunstein <[EMAIL PROTECTED]> writes:
>
> | I don't see the point really.
>
> Depends on how anal we want to be.
> (and no-path is a bit more secure than path)
Of course. But we still have some uses of system... so I leave you the
interpretation of how "ana
Alfredo Braunstein <[EMAIL PROTECTED]> writes:
| I don't see the point really.
Depends on how anal we want to be.
(and no-path is a bit more secure than path)
--
Lgb
Lars Gullik BjÃnnes wrote:
> Is execvp really nice?
>
> I would have prefered execv since malicious programs can then not be
> called from arbitrary locations (in the PATH).
I don't see this as a problem, as you cannot even specify the name of the
program...
the attacker would have to introduce
Alfredo Braunstein <[EMAIL PROTECTED]> writes:
| Can someone test the current patch with documents with external insets to
| see if all works OK?
>
| Attached also a testcase showing the bad behaviour before the patch.
>
| Should we do something equivalent for converters?
>
| I don't know if playi
Can someone test the current patch with documents with external insets to
see if all works OK?
Attached also a testcase showing the bad behaviour before the patch.
Should we do something equivalent for converters?
I don't know if playing with the file name to be converted you can also
execute ar
18 matches
Mail list logo
