bug#22266: [libparted] jump instruction and boot code is corrupted with random bytes after fat is resized

On Tue, Dec 29, 2015 at 06:08:31PM +0800, Tom Yan wrote:
> https://gist.github.com/tomty89/2a1ca8698bad5ae969ad/revisions?diff=split
> 
> The (last) diff compares a FAT32 resized with a working commit of
> libparted and a not working one. You can see that in the latter, the
> jump instruction and boot code in the boot sectors are not preserved
> but completely corrupted by random bytes. (FWIW, the first byte "eb"
> is crucial for Windows to recognize the filesystem)
> 
> The last working commit I can confirm is:
> http://git.savannah.gnu.org/cgit/parted.git/commit/?id=d0a4cc1b57750a92afb48b229e4791154afa322b
> 
> which is the commit right before:
> http://git.savannah.gnu.org/cgit/parted.git/commit/?id=80678bdd957cf49a9ccfc8b88ba3fb8b4c63fc12
> 
> where FAT resizing no longer works, until:
> http://git.savannah.gnu.org/cgit/parted.git/commit/?id=5adae27101565a5d6fed4aadf28ddb39872e41f5
> 
> which is the first commit I can confirm to have this issue.
> 
> I've also tested the latest commit:
> http://git.savannah.gnu.org/cgit/parted.git/commit/?id=388bab890a4e09b09d2428c0e773ed083295f91b
> 
> which does not work properly either.
> 
> See this bug report for more details:
> https://bugzilla.gnome.org/show_bug.cgi?id=759916

So you are saying that commit 5adae27101565a5d6fed4aadf28ddb39872e41f5
does *not* fix the problem?

We should drop this resize code completely. I still think it was a
mistake to revive it as a library. Filesystems should be managed by the
upstream filesystem code, not parted.

-- 
Brian C. Lane | Anaconda Team | IRC: bcl #anaconda | Port Orchard, WA (PST8PDT)