Re: PATCH: cdrecord: avoiding scsi device numbering for ide devices

[Joerg Schilling]

	---->	This is a resend as it seems that somebody did remove
		linux-kernel@xxxxxxxxxxxxxxx

>From: "H.Rosmanith (Kernel Mailing List)" <kernel@xxxxxxxxxxxxxxxxx>

>> I can just that if you set 64KiB now, it'll be a much better alround
>> value.

>very good, it works with 64kB:

>    887         /*
>    888          * First try to raise the DMA limit to a moderate value that
>    889          * most likely does not use up all kernel memory.
>    890          */
>    891         //val = 126*1024;
>    892         val = 64*1024;

>now cdrecord will happily use the scsi-linux-sg driver even for the IDE
>burners connected to the "Digitus" controllers handled by the siimage driver.

So you found a nasty bug in the Linux kernel :-(

The official behavior for SG_SET_RESERVED_SIZE is:

1)	You call SG_SET_RESERVED_SIZE with whatever size val you like

2)	It says "thank you" and _always_ returns success

3)	You call SG_GET_RESERVED_SIZE to read the value set up
	in the kernel. This value is not directly related to the value
	used with SG_SET_RESERVED_SIZE. The proposal from Douglas
	Gilbert was to always use 512 KB. I am unhappy with this behavior
	but I have to take it as it has been defined by the Author :-(
	He told me that the value returned with SG_GET_RESERVED_SIZE
	is the value available for DMA and may be much smaller than
	the size used with SG_SET_RESERVED_SIZE.

AGAIN: I am unhappy with this behavior but this _is_ the official behavior.
If the current Linux kernel for some reasons does not behave this way it
is broken and needs to be fixed.

See also: http://www.mail-archive.com/cdwrite@xxxxxxxxxxxxxxxx/msg00232.html

Jörg

-- 
 EMail:joerg@xxxxxxxxxxxxxxxxxxxxxxxxxxx (home) Jörg Schilling D-13353 Berlin
       js@xxxxxxxxxxxxxxx		(uni)  If you don't have iso-8859-1
       schilling@xxxxxxxxxxxxxxxxxxx	(work) chars I am J"org Schilling
 URL:  http://www.fokus.fraunhofer.de/usr/schilling ftp://ftp.berlios.de/pub/schily
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/