Re: [PATCH V7 1/3] genalloc:support memory-allocation with bytes-alignment to genalloc

[Scott Wood <scottwood@freescale.com>]

On Wed, 2015-09-09 at 21:26 -0500, Zhao Qiang-B45475 wrote:
> On Wed, 2015-09-10 at 12:38AM -0500, Wood Scott-B07421 wrote:
> > -----Original Message-----
> > From: Wood Scott-B07421
> > Sent: Thursday, September 10, 2015 12:38 AM
> > To: Zhao Qiang-B45475
> > Cc: linux-kernel@vger.kernel.org; linuxppc-dev@lists.ozlabs.org;
> > lauraa@codeaurora.org; Xie Xiaobo-R63061; benh@kernel.crashing.org; Li
> > Yang-Leo-R58472; paulus@samba.org
> > Subject: Re: [PATCH V7 1/3] genalloc:support memory-allocation with
> > bytes-alignment to genalloc
> > 
> > On Sat, 2015-09-05 at 22:13 -0500, Zhao Qiang-B45475 wrote:
> > > On Wed, 2015-09-02 at 10:18AM +0800, Wood Scott-B07421 wrote:
> > > > -----Original Message-----
> > > > From: Wood Scott-B07421
> > > > Sent: Wednesday, September 02, 2015 10:18 AM
> > > > To: Zhao Qiang-B45475
> > > > Cc: linux-kernel@vger.kernel.org; linuxppc-dev@lists.ozlabs.org;
> > > > lauraa@codeaurora.org; Xie Xiaobo-R63061; benh@kernel.crashing.org;
> > > > Li Yang-Leo-R58472; paulus@samba.org
> > > > Subject: Re: [PATCH V7 1/3] genalloc:support memory-allocation with
> > > > bytes-alignment to genalloc
> > > > 
> > > > On Tue, 2015-09-01 at 21:10 -0500, Zhao Qiang-B45475 wrote:
> > > > > On Wed, 2015-09-02 at 08:38AM +0800, Wood Scott-B07421 wrote:
> > > > > > -----Original Message-----
> > > > > > From: Wood Scott-B07421
> > > > > > Sent: Wednesday, September 02, 2015 8:30 AM
> > > > > > To: Zhao Qiang-B45475
> > > > > > Cc: linux-kernel@vger.kernel.org; linuxppc-dev@lists.ozlabs.org;
> > > > > > lauraa@codeaurora.org; Xie Xiaobo-R63061;
> > > > > > benh@kernel.crashing.org; Li Yang-Leo-R58472; paulus@samba.org
> > > > > > Subject: Re: [PATCH V7 1/3] genalloc:support memory-allocation
> > > > > > with bytes-alignment to genalloc
> > > > > > 
> > > > > > On Mon, 2015-08-31 at 16:58 +0800, Zhao Qiang wrote:
> > > > > > > Bytes alignment is required to manage some special RAM, so add
> > > > > > > gen_pool_first_fit_align to genalloc, meanwhile add
> > > > > > > gen_pool_alloc_data to pass data to
> > > > > > > gen_pool_first_fit_align(modify gen_pool_alloc as a wrapper)
> > > > > > > 
> > > > > > > Signed-off-by: Zhao Qiang <qiang.zhao@freescale.com>
> > > > > > > ---
> > > > > > > Changes for v6:
> > > > > > >       - patches set v6 include a new patch because of using
> > > > > > >       - genalloc to manage QE MURAM, patch 0001 is the new
> > > > > > >       - patch, adding bytes alignment for allocation for use.
> > > > > > > Changes for v7:
> > > > > > >       - cpm muram also need to use genalloc to manage, it has
> > > > > > >         a function to reserve a specific region of muram,
> > > > > > >         add offset to genpool_data for start addr to be
> > allocated.
> > > > > > 
> > > > > > This seems to be describing more than just the changes in this
> > patch.
> > > > > > What does also handling cpm have to do with this patch?  Are you
> > > > > > adding support for reserving a specific region in this patch?  I
> > > > > > don't see it, and in any case it should go in a different patch.
> > > > > 
> > > > > Yes, I added. The code below can support the function.
> > > > >       offset_bit = (alignment->offset + (1UL << order) - 1) >>
> > order;
> > > > >       return bitmap_find_next_zero_area(map, size, start +
> > > > > offset_bit,
> > > > nr,
> > > > >                         align_mask);
> > > > > 
> > > > > CPM has an function cpm_muram_alloc_fixed, needing to allocate
> > > > > muram from a Specific offset. So I add the code and add offset to
> > struct data.
> > > > 
> > > > I thought the offset was related to the previous discussion of
> > > > checking for allocation failure.  Are you using it to implement
> > > > alloc_fixed()?  If so, please don't.  Besides the awkward
> > > > implementation (what does it logically have to do with
> > > > gen_pool_first_fit_align?), it does not appear to be correct -
> > > > - what happens with multiple chunks?  What happens if part of the
> > > > region the caller is trying to reserve is already taken?  Implement
> > > > a proper function to reserve a fixed genalloc region.
> > > > 
> > > The offset is which allocation block address need to be larger than,
> > > Not equal to, it really like the parameter start of the algo(the
> > > bitnumber To start searching at).
> > 
> > cpm_muram_alloc_fixed() is not "search starting at this offset".  It is
> > "reserve this exact range or fail".
> 
> Yes, you are right! How about to add a new algo into genalloc to search 
> At offset, then handle it in muram layer, if the address return from 
> genalloc
> Is not equal to offset, return negative number?

If you're adding a new algorithm, why not make it actually do what you want 
rather than adding something different and fixing it up in the caller?

-Scott