Re: [PATCH 1/6] firewire: core: reduce stack usage in bus reset tasklet

[Stefan Richter]

Stefan Richter wrote:
> pageexec@xxxxxxxxxxx wrote:
> > On 6 Sep 2009 at 18:48, Stefan Richter wrote:
> >
> > > Index: linux-2.6.31-rc9/drivers/firewire/core-card.c
> > > ===================================================================
> > > --- linux-2.6.31-rc9.orig/drivers/firewire/core-card.c
> > > +++ linux-2.6.31-rc9/drivers/firewire/core-card.c
> > > @@ -38,16 +38,21 @@
> > >  
> > >  #include "core.h"
> > >  
> > > -int fw_compute_block_crc(u32 *block)
> > > +int fw_compute_block_crc(u32 *block, gfp_t flags)
> > >  {
> > > -    __be32 be32_block[256];
> > > -    int i, length;
> > > +    static __be32 *be32_block;
> >        ^^^^^^
> >
> > did you actually mean that to be static? if so, then you might as well 
> > allocate the
> > buffer statically and not worry about a runtime allocation failure.
>
> Uh, that's a cut'n'paste mistake.  It shouldn't be static.  Thanks, I'll 
> correct that before I put it into linux1394-2.6.git.
>
> > > +    int i, length = (*block >> 16) & 0xff;
> > > +
> > > +    be32_block = kmalloc(length * 4, flags);
> > > +    if (WARN_ON(!be32_block))
> > > +        goto out;
> > >  
> > > -    length = (*block >> 16) & 0xff;
> > >      for (i = 0; i < length; i++)
> > >          be32_block[i] = cpu_to_be32(block[i + 1]);
> > >      *block |= crc_itu_t(0, (u8 *) be32_block, length * 4);
> > >  
> > > +    kfree(be32_block);
> > > + out:
> > >      return length;
> > >  }

This can be optimized further to get rid of the temporary be32_block 
entirely.

There are two users of fw_compute_block_crc:  The functions which 
generate the local Config ROM CSR and then pass it down to the low-level 
driver, and the function which which generates the local Topology Map 
CSR and stores it for core-transaction.c::handle_topology_map() to read 
on demand.

The low-level driver then needs to convert the Config ROM to __be32[]. 
Likewise, handle_topology_map needs to convert the Topology Map to 
__be32[].  They would be both better off if we stored their input as 
__be32[] in the first place, and of course do so before we compute the 
CRC.  Duh.

I'll rewrite patch 1/6 and 5/6 accordingly.
--
Stefan Richter
-=====-==--= =--= --===
http://arcgraph.de/sr/
--
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/