Mailinglist Archive: packet-writing (59 mails)

< Previous Next >
Re: [packet-writing] Re: [release] packet-0.0.2c
  • From: Jens Axboe <axboe@xxxxxxx>
  • Date: Sat, 9 Sep 2000 14:46:57 +0000 (UTC)
  • Message-id: <20000909164444.J462@xxxxxxx>
On Sat, Sep 09 2000, adam@xxxxxxxxxxxx wrote:
> SCSI-over-IDE
> (EIDE CDRW drive)
>
> SMP

Ok, I test on raw atapi and scsi on my test machine. Also SMP.

> > All the data or some of it? Could I get you to find out where the
> > corruption is taking place -- try doing a cmp of some corrupted files
> > and the originals, to see where the offset is.
>
> [root@pepsi cd2]# cmp ~adam/ncd19r.tar.gz ncd19r.tar.gz
> /home/adam/ncd19r.tar.gz ncd19r.tar.gz differ: char 956417, line 2467

Same deal, corruption starts at the beginning of a new 2k sized buffer.
Could you read the first 2048 bytes from the CD-ROM file and check that
this buffer is zero? It is here, which means that a gather-block wasn't
brought properly into memory before being written out.

--
* Jens Axboe <axboe@xxxxxxx>
* SuSE Labs

< Previous Next >
Follow Ups
References