On Sun, Mar 09, 2003 at 12:57:11PM -0700, Bob Proulx wrote:
> Osamu Aoki wrote:
> > My post on this thread needs few errata:
> > 1. CD image size can be obtained by "mount" the CD and run "df".
>
> Interesting. Thanks for educating us about that.
>
> > 2. Only "readcd" run with exact CD size can extract the correct image
> > 3. "dd" tends to copy most data correct but loses last few KiB of data.
>
> I have been reading this thread with great interest because I had
> determined the following behavior.
>
> cdrecord image-in.iso
> dd if=/dev/cdrw of=image-out-cdrw.iso # works
> cmp image-in.iso image-out-cdrw.iso # passes
>
> dd if=/dev/cdrom of=image-out-cdrom.iso # reports errors on last few kb
> cmp image-in.iso image-outcdrom.iso # fails
Are you sure this is really just HW difference? See below.
> The images compare exactly when being read from the cdrw drive. But
> if I move the CD out of the writer that was used to write the image
> and put it in a different CD reader only, not a writer, then the
> repeat the dd then the image I retrieve is different and fails
> compare. The last few kb of data is missing. The dd reports an I/O
> error when trying to read those bytes on a cd reader only device. It
> was fine on a cd writer device.
...
OK, this prompted me to do more research. It looks somewhat complicated
and it may be somewhat firmware dependent error handling issues
involved. "cdtest" is a simple script shown below and "dd" is showing
different result once error occurs. This memory effect erases once CD
are ejected.
---[cdtest script]---
#!/bin/sh
dd if=/dev/cdrom of=sarge-${1}-$((${1}/2048)) bs=2048 count=$((${1}/2048))
ls -l sarge-${1}-$((${1}/2048))
---
[EMAIL PROTECTED]:root# cdtest $((46301184))
dd: reading `/dev/cdrom': Input/output error
22564+0 records in
22564+0 records out
46211072 bytes transferred in 48.380704 seconds (955155 bytes/sec)
-rw-r--r-- 1 root root 46211072 Mar 9 16:54 sarge-46301184-22608
[EMAIL PROTECTED]:root# cdtest $((46301184))
22608+0 records in
22608+0 records out
46301184 bytes transferred in 53.311948 seconds (868495 bytes/sec)
-rw-r--r-- 1 root root 46301184 Mar 9 16:55 sarge-46301184-22608
[EMAIL PROTECTED]:root# dd if=/dev/cdrom of=sarge-free-2ndhand
90436+0 records in
90436+0 records out
46303232 bytes transferred in 43.151651 seconds (1073035 bytes/sec)
[EMAIL PROTECTED]:root# dd if=sarge-free-2ndhand of=sarge-free-46301184 bs=46301184
count=1
1+0 records in
1+0 records out
46301184 bytes transferred in 0.295885 seconds (156483718 bytes/sec)
(... few more error free run of cdtest)
[EMAIL PROTECTED]:root# ls -l sarge*
-rw-r--r-- 1 root root 46301184 Mar 9 16:55 sarge-46301184-22608
-rw-r--r-- 1 root root 46301184 Mar 9 16:49 sarge-46302208-22608
-rw-r--r-- 1 root root 46303232 Mar 9 16:47 sarge-46303232-22609
-rw-r--r-- 1 root root 46303232 Mar 9 16:52 sarge-46401184-22656
-rw-r--r-- 1 root root 46303232 Mar 9 17:00 sarge-free-2ndhand
-rw-r--r-- 1 root root 46301184 Mar 9 17:10 sarge-free-46301184
[EMAIL PROTECTED]:root# for xx in sarge* ; do cat $xx | md5sum ; done
65a9e488ddc9a9453685ca496af6e242 -
65a9e488ddc9a9453685ca496af6e242 -
da940de4ad9a634b72e7c899322fd8a3 -
da940de4ad9a634b72e7c899322fd8a3 -
da940de4ad9a634b72e7c899322fd8a3 -
65a9e488ddc9a9453685ca496af6e242 -
As you can see, once error free "dd" are established, "dd" return
correct data with some (2048 for my drive but I think this should be HW
dependent.)
When "dd" errors, the data size seems significantly smaller. May be it
drops entire data cache (either firmware level or Linux block read
level, I have no idea.).
So now I know, dd can be used to read image correctly once after CD are
read with error and never ejected. Interesting.
--
~\^o^/~~~ ~\^.^/~~~ ~\^*^/~~~ ~\^_^/~~~ ~\^+^/~~~ ~\^:^/~~~ ~\^v^/~~~ +++++
Osamu Aoki <[EMAIL PROTECTED]> Cupertino CA USA, GPG-key: A8061F32
.''`. Debian Reference: post-installation user's guide for non-developers
: :' : http://qref.sf.net and http://people.debian.org/~osamu
`. `' "Our Priorities are Our Users and Free Software" --- Social Contract
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
