Linux ISO-9660 Rock Ridge bug needs fix
A message that is not in the archive (sent off-list or not yet received): [email protected]
A message that is not in the archive (sent off-list or not yet received): [email protected]
Review
Quoted 6 lines · Luca Tettamanti
+++
b/fs/isofs/rock.c(quoted diff) +4
-0+ + /* Rock Ridge V1.12, override inode number */ + if (rr->len == 44) + inode->i_ino = isonum_733(rr->u.PX.inode);
I think it's wrong again, it will break as soon as rockridge 1.13 defines an additional field. s,==,>=, should do the trick.
BTW and without digging in the code: Since version 2 will be binary incompatible, is there a version check?
BTW2, Just to be cautionous: what will happen if somebody forces the same inode number on two different entries?
Signature · 4 lines
Ich danke GMX dafür, die Verwendung meiner Adressen mittels per SPF verbreiteten Lügen zu sabotieren. http://david.woodhou.se/why-not-spf.html
Quoted 9 lines · Bodo Eggert
Review
Quoted 6 lines · Luca Tettamanti
+++
b/fs/isofs/rock.c(quoted diff) +4
-0+ + /* Rock Ridge V1.12, override inode number */ + if (rr->len == 44) + inode->i_ino = isonum_733(rr->u.PX.inode);
Quoted 5 lines
I think it's wrong again, it will break as soon as rockridge 1.13 defines an additional field. s,==,>=, should do the trick.
BTW and without digging in the code: Since version 2 will be binary incompatible, is there a version check?
There is de-facto no consistency check on Linux.
It would be easy to create special ISO images that cause Linux to panic from many other places in the ISO-9660 code.
Quoted 2 lines
BTW2, Just to be cautionous: what will happen if somebody forces the same inode number on two different entries?
What happenes when somebody set the traffic light to "green" in all directions?
What happenes if someone sets the same inode number for different files on UDF?
This is something you cannot check.
Jörg
Signature · 4 lines
EMail:[email protected] (home) Jörg Schilling D-13353 Berlin [email protected] (uni) [email protected] (work) Blog: http://schily.blogspot.com/ URL: http://cdrecord.berlios.de/old/private/ ftp://ftp.berlios.de/pub/schily
Quoted 4 lines · Joerg Schilling
Quoted 2 lines · Bodo Eggert
BTW2, Just to be cautionous: what will happen if somebody forces the same inode number on two different entries?
[...]
Quoted 1 line
This is something you cannot check.
Exactly that's why I'd ignore the on-disk "inode number" and instead use the generated one untill someone comes along with a clever idea to fix the issue or can show that it's mostly hermless.
Signature · 1 line
Artificial Intelligence usually beats real stupidity.
Quoted 6 lines · Bodo Eggert
Quoted 4 lines · Joerg Schilling
Quoted 2 lines · Bodo Eggert
BTW2, Just to be cautionous: what will happen if somebody forces the same inode number on two different entries?
Quoted 6 lines
[...]
Quoted 1 line
This is something you cannot check.
Exactly that's why I'd ignore the on-disk "inode number" and instead use the generated one untill someone comes along with a clever idea to fix the issue or can show that it's mostly hermless.
I could understand you in case that Linux would do some basic consistency checks in the iso-9660 code already.....
Show me another program besides mkisofs that implements inode numbers _and_ does it wrong.
Jörg
Signature · 4 lines
EMail:[email protected] (home) Jörg Schilling D-13353 Berlin [email protected] (uni) [email protected] (work) Blog: http://schily.blogspot.com/ URL: http://cdrecord.berlios.de/old/private/ ftp://ftp.berlios.de/pub/schily
Quoted 8 lines · Joerg Schilling
Quoted 7 lines · Bodo Eggert
Quoted 4 lines · Joerg Schilling
Quoted 2 lines · Bodo Eggert
BTW2, Just to be cautionous: what will happen if somebody forces the same inode number on two different entries?
[...]
Quoted 8 lines
Quoted 5 lines
Quoted 1 line
This is something you cannot check.
Exactly that's why I'd ignore the on-disk "inode number" and instead use the generated one untill someone comes along with a clever idea to fix the issue or can show that it's mostly hermless.
I could understand you in case that Linux would do some basic consistency checks in the iso-9660 code already.....
ISO9660-over-NFS is no big usecase, linked files on CD aren't common and the current code just won't benefit from hardlinks. Therefore I don't see a compelling reason to use this feature without first thinking about possible attacks.
Even if there are more holes to be plugged, punching even more holes doesn't help. Instead, we should hope that someone finds the time to plug these holes and praise him when he comes.
Quoted 2 lines
Show me another program besides mkisofs that implements inode numbers _and_ does it wrong.
My hacked mkisofs, which I'll use to r00t you all and gain world domination.
Signature · 2 lines
"Don't draw fire; it irritates the people around you." -Your Buddies
Quoted 11 lines · Bodo Eggert
Quoted 6 lines
Quoted 3 lines
Exactly that's why I'd ignore the on-disk "inode number" and instead use the generated one untill someone comes along with a clever idea to fix the issue or can show that it's mostly hermless.
I could understand you in case that Linux would do some basic consistency checks in the iso-9660 code already.....
ISO9660-over-NFS is no big usecase, linked files on CD aren't common and the current code just won't benefit from hardlinks. Therefore I don't see a compelling reason to use this feature without first thinking about possible attacks.
It looks like you did never really think about the problem or that you just don't know what you are talking about :-(
A typical UNIX standard installation has more than 3000 hard linked file name entries in the root filesystem. There are many people who like to backup their root filesystem via mkisofs and all life CDs and install CDs depend on hard linked files.
But this seems to be a frequent way of dealing with implementation deficits:
define the problem to be not existant....
Quoted 3 lines
Even if there are more holes to be plugged, punching even more holes doesn't help. Instead, we should hope that someone finds the time to plug these holes and praise him when he comes.
It seems that you did not think about the problem...
Wrong inode numbes do not open holes, they just create a garbage in garbage out behavior, but they will not cause the OS to panic.
The unfixed bugs in the Linux iso-9660 implementation on the other side allow me to create a CD that causes Linux to panic without taking me a long time.
Quoted 4 lines
Quoted 2 lines
Show me another program besides mkisofs that implements inode numbers _and_ does it wrong.
My hacked mkisofs, which I'll use to r00t you all and gain world domination.
This is just a junk statement, describe what you believe than can be done...
Jörg
Signature · 4 lines
EMail:[email protected] (home) Jörg Schilling D-13353 Berlin [email protected] (uni) [email protected] (work) Blog: http://schily.blogspot.com/ URL: http://cdrecord.berlios.de/old/private/ ftp://ftp.berlios.de/pub/schily
Ar Iau, 2006-10-19 am 23:39 +0200, ysgrifennodd Joerg Schilling:
Quoted 2 lines
Wrong inode numbes do not open holes, they just create a garbage in garbage out behavior, but they will not cause the OS to panic.
Thats implementation dependant as people who stress OS environments and do fuzz testing have provided again and again and again - including a previous Linux over ext3 case. There isn't exactly any hurry - its evident from the lack of reports that you are the first tool author whose beta tool even uses this feature.
Quoted 2 lines
The unfixed bugs in the Linux iso-9660 implementation on the other side allow me to create a CD that causes Linux to panic without taking me a long time.
And those should indeed be fixed.