Re: [RFT] Support for ~2144 SCSI discs

[Richard Gooch <rgooch@ras.ucalgary.ca>]

Jeremy Higdon writes:
> With the sard patch and a 64 bit system, you start having
> trouble at around 103 configured disks, because of the following

So even without my patch, sard doesn't support the previous limit of
128 devices.

> line in sd_init() (sd.c), because kmalloc doesn't like allocating
> large chunks of memory:
> 
>         sd = kmalloc((sd_template.dev_max << 4) *
>                                           sizeof(struct hd_struct),
>                                           GFP_ATOMIC);
> 
> Without sard, you'd have problems past 512 disks.

Yes, when I was coding up the patch I noticed the use of GFP_ATOMIC in
the allocation calls. I have two questions:
- can we use GFP_KERNEL instead (why use GFP_ATOMIC)
- can we switch to vmalloc() instead of kmalloc()?

> With the sard patch, the hd_struct looks like the following:
> 
> struct hd_struct {
>         long start_sect;
>         long nr_sects;
>         devfs_handle_t de;              /* primary (master) devfs entry  */
> 
>         int number;                     /* stupid old code wastes space  */
> 
>         /* Performance stats: */
>         unsigned int ios_in_flight;
>         unsigned int io_ticks;
>         unsigned int last_idle_time;
>         unsigned int last_queue_change;
>         unsigned int aveq;
> 
>         unsigned int rd_ios;
>         unsigned int rd_merges;
>         unsigned int rd_ticks;
>         unsigned int rd_sectors;
>         unsigned int wr_ios;
>         unsigned int wr_merges;
>         unsigned int wr_ticks;
>         unsigned int wr_sectors;
> };
> 
> The caveat is that I'm looking at a patch that is a few months old (I
> couldn't find where the latest version of the kernel patch is).

Do the stats have to be kept on a per-partition basic? What about
per-device instead?

				Regards,

					Richard....
Permanent: rgooch@atnf.csiro.au
Current:   rgooch@ras.ucalgary.ca