[vox-tech] gzip bug?

Peter Jay Salzman vox-tech@lists.lugod.org
Thu, 14 Mar 2002 13:36:38 -0800


hmmm... i'm still reluctant to say that there's a bug in gzip or bzip2.
it would be like saying that mcdonalds has been making french fries
incorrectly all these years.  it's like, someone should've noticed by
now.

but i'm fresh out of ideas.  i'm not sure what else to say.

i took a look at deja just now and didn't find anything particularly
useful for your situation.  why don't you file that bug report and see
what they have to say?

pete

begin Kevin Dawson <kdawson@ucdavis.edu> 
> Hi Pete,
> 
> The files I'm having the problem with are just 50M and 30M after gzip.
> 
> I followed your suggestion. Everything was going fine up to 1M. Using
> 10M, I had to learn that I didn't have 20G disk space available on
> /home. Using 100k, the created file size is in the range of my problem
> files (50M), can be compressed to only 50k, and uncompressed perfectly.
> It looks like that the problem is dependent on the size of the
> compressed file (30M in my case).
> 
> bunzip2 is also having trouble. It says:
> bunzip2: Data integrity error when decompressing.
> 	Input file=bigfile.bz2, output file=bigfile
> 
> It is possible that the compressed file(s) have become corrupted.
> You can use the -tvv option to test integrity of such files.
> 
> When I do the test, it will print
> [N: huff+mtf rt+rld] whereas N goes up to 29.
> Then it prints: data integrity (CRC) error in data.
> 
> When gunzipping the files, I don't hear any unusual disk churning.
> 
> Thanks again,
> 
> Kevin
> 
> 
> 
> 
> 
> 
> -----Original Message-----
> From: vox-tech-admin@lists.lugod.org
> [mailto:vox-tech-admin@lists.lugod.org] On Behalf Of Peter Jay Salzman
> Sent: Thursday, March 14, 2002 10:38 AM
> To: vox-tech@lists.lugod.org
> Subject: Re: [vox-tech] gzip bug?
> 
> hi kevin,
> 
> just out of curiosity, when you say it happens with all big files, how
> big is big?  1k?  10k?  100k?  are youa bioinformatics dude?  ;-)  what
> i'm getting at is -- are you dealing with files near the 2GB limit?
> 
> what happens when you try to do something like:
> 
> dd if=/dev/zero of=mybigfile count=10k
> cp mybigfile myotherbigfile
> gzip mybigfile
> gzip -d mybigfile
> diff mybigfile.gz myotherbigfile
> 
> change the 10k to whatever you need to reproduce the crc error.  also,
> 10k of /dev/zero should produce about a mp3 sized file (5 Megs).
> 
> if you can reproduce the CRC error this way, i totally encourage you to
> write to bug-gzip@gnu.org.   gzip is one of the things that really ought
> to be rock solid.  you can screw around with USB and TV tuner cards, but
> gzip and bzip2 are our meat and potatoes.
> 
> btw, does this problem happen with bzip2?
> 
> also, when you gunzip these large files, how hard does your machine
> thrash?  does the hard drive go like mad?
> 
> pete
> 
> begin Kevin Dawson <kdawson@ucdavis.edu> 
> > Hi Pete,
> > 
> > Thank you for your suggestion. I'm using gzip 1.3 and gunzip 1.3 on a
> > VFS file system. I'm having problems with practically all big files.
> The
> > files are corrupted without ever leaving the Linux machine. For
> > trouble-shooting with WinZip, I transferred some files to Windows
> using
> > Samba.
> > 
> > The exact error message reads as:
> > gunzip: MX.mat.gz: invalid compressed data--crc error
> > 
> > Your filter suggestion is great. It works, although, I'm still getting
> > the error message:
> > gzip: stdin: invalid compress data--crc error
> > 
> > Thank you,
> > Kevin
> > 
> > 
> > -----Original Message-----
> > From: vox-tech-admin@lists.lugod.org
> > [mailto:vox-tech-admin@lists.lugod.org] On Behalf Of Peter Jay Salzman
> > Sent: Thursday, March 14, 2002 9:59 AM
> > To: vox-tech@lists.lugod.org
> > Subject: Re: [vox-tech] gzip bug?
> > 
> > hi kevin,
> > 
> > stream of consciousness:
> > 
> > are you saying ALL files on redhat can't be gunzipped?  or only
> > particular files?  was this file transfered using windows ftp or http?
> > perhaps it went through a NL/CR conversion?  faulty kermit transfer?
> > where was the file created?  under a vfat system or under a VFS
> (linux)
> > system?
> > 
> > try to use gzip as a filter:
> > 
> > gzip -d <myfile.gz >myfile
> > 
> > this will unzip as far as gzip can go before reaching the CRC error.
> > it might recover some of your data.
> > 
> > also, can you post the exact error message (always a good thing to do
> > when asking for help when something complains!)
> > 
> > can you post gzip's version number and mail me (not the list) a copy
> of
> > the file you're trying to unzip?  i'll give it go here.
> > 
> > pete
> > 
> > 
> > begin Kevin Dawson <kdawson@ucdavis.edu> 
> > > Hi,
> > > 
> > > I cannot gunzip my gzipped files on RedHat 7.1. I get CRC error
> > > messages. On the other hand, when the files are transferred to
> > Windows,
> > > WinZip opens them without any complain. Do you know any known bugs
> in
> > > gzip or gunzip? Do you know another uncompressing utility on Linux
> > which
> > > does not invoke RedHat's gunzip to uncompress a .gz file.
> > > 
> > > Any suggestion is appreciated.
> > > 
> > > Kevin
> > > 
> > > _______________________________________________
> > > vox-tech mailing list
> > > vox-tech@lists.lugod.org
> > > http://lists.lugod.org/mailman/listinfo/vox-tech
> > _______________________________________________
> > vox-tech mailing list
> > vox-tech@lists.lugod.org
> > http://lists.lugod.org/mailman/listinfo/vox-tech
> > 
> > _______________________________________________
> > vox-tech mailing list
> > vox-tech@lists.lugod.org
> > http://lists.lugod.org/mailman/listinfo/vox-tech
> _______________________________________________
> vox-tech mailing list
> vox-tech@lists.lugod.org
> http://lists.lugod.org/mailman/listinfo/vox-tech
> 
> _______________________________________________
> vox-tech mailing list
> vox-tech@lists.lugod.org
> http://lists.lugod.org/mailman/listinfo/vox-tech