[U-Boot] [PATCH] mmc:dcache: Cache line size aligned internal MMC buffers

[Anton Staaf]

On Mon, Aug 22, 2011 at 11:31 AM, Mike Frysinger <vapier at gentoo.org> wrote:
> On Monday, August 22, 2011 14:15:22 Anton Staaf wrote:
>
> please dont top post

Sorry about that.

>> Yes, this would be a much preferable approach.  The only problem that
>> I encountered when attempting to do that in the Chromium U-Boot was
>> that there are exposed stand alone U-Boot API's that can pass in
>> pointers that make their way all the way down to the device driver.
>> Two solutions occurred to me.  One was to put an Error or assert as
>> high up as possible and expect violators to fix their code, the other
>> was to patch things up in lower layers (right now we have a bounce
>> buffer in the tegra MMC driver to deal with unaligned buffers).  My
>> intent was that this would give us a functional U-Boot while we took
>> the time to clean up as many of the unaligned buffers that U-Boot
>> uses, and then tackle the question of stand alone applications.
>
> could you elaborate on the code paths in question ?

Heh, now I've got to go fine them again.  The ub_dev_read function in
the example/api/glue.c is an example of a use of the public syscall
API that as far as I can tell passes down the buffer pointer all the
way to the driver layer.

As for the unaligned buffers I've run into:

fs/ext2/dev.c: sec_buf is allocated on the stack and thus is not
guaranteed to be aligned.
drivers/mmc/mmc.c: ext_csd in mmc_change_freq is allocated on the stack.
drivers/mmc/mmc.c: scr and switch_status in sd_change_freq are
allocated on the stack.
drivers/mmc/mmc.c: ext_csd in mmc_startup is allocated on the stack.
disk/part_efi.c: gpt_header in print_part_efi is allocated on the stack.
disk/part_efi.c: gpt_header in get_partition_info_efi is allocated on the stack.
disk/part_efi.c: legacy_mbr in test_part_efi is allocated on the stack.
disk/part_efi.c: pte in alloc_read_gpt_entries is allocated using
malloc (not memalign).

I have working solutions to all of these but they generally use the
lazy allocation of an aligned buffer as needed pattern.  I wanted to
see the results of Lukasz current thread before sending them upstream.
 But that's probably not right.  I should probably send them now so we
can see a number of examples of the problem and get a better idea of
how to fix it.

Thanks,
    Anton

> -mike
>