Re: [patch] SGIIOC4 limit request size

[Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>]

On 01 Feb 2006 03:59:16 -0500, Jes Sorensen <jes@sgi.com> wrote:
> Hi,

Hi,

> This one takes care of a problem with the SGI IOC4 driver where it
> hits DMA problems if the request grows too large.

Does this happen only for CONFIG_IA64_PAGE_SIZE_4KB=y
or CONFIG_IA64_PAGE_SIZE_8KB=y?

from sgiioc4.c:

/* Each Physical Region Descriptor Entry size is 16 bytes (2 * 64 bits) */
/* IOC4 has only 1 IDE channel */
#define IOC4_PRD_BYTES       16
#define IOC4_PRD_ENTRIES     (PAGE_SIZE /(4*IOC4_PRD_BYTES))

As limiting request size to 127 sectors punishes performance
wouldn't it be better to define IOC4_PRD_ENTRIES to 256
if this is possible (would need 4 pages for PAGE_SIZE=4096
and 2 for PAGE_SIZE=8192)?

Cheers.
Bartlomiej

> Cheers,
> Jes
>
> Avoid requests larger than the number of SG table entries, to avoid
> DMA timeouts.
>
> Signed-off-by: Jes Sorensen <jes@sgi.com>
>
> ----
>
>  drivers/ide/pci/sgiioc4.c |    8 +++++++-
>  1 files changed, 7 insertions(+), 1 deletion(-)
>
> Index: linux-2.6/drivers/ide/pci/sgiioc4.c
> ===================================================================
> --- linux-2.6.orig/drivers/ide/pci/sgiioc4.c
> +++ linux-2.6/drivers/ide/pci/sgiioc4.c
> @@ -1,5 +1,5 @@
>  /*
> - * Copyright (c) 2003 Silicon Graphics, Inc.  All Rights Reserved.
> + * Copyright (C) 2003, 2006 Silicon Graphics, Inc.  All Rights Reserved.
>   *
>   * This program is free software; you can redistribute it and/or modify it
>   * under the terms of version 2 of the GNU General Public License
> @@ -613,6 +613,12 @@
>         hwif->ide_dma_lostirq = &sgiioc4_ide_dma_lostirq;
>         hwif->ide_dma_timeout = &__ide_dma_timeout;
>         hwif->INB = &sgiioc4_INB;
> +
> +       /*
> +        * Limit the request size to avoid DMA timeouts when
> +        * requesting  more entries than goes in the sg table.
> +        */
> +       hwif->rqsize = 127;
>  }
>
>  static int __devinit
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/