Re: [PATCH v2 3/3] zram: pass gfp from zcomp frontend to backend

[Minchan Kim <minchan@kernel.org>]

Hi Sergey,

On Wed, Nov 25, 2015 at 09:46:47PM +0900, Sergey Senozhatsky wrote:
> On (11/25/15 14:51), Minchan Kim wrote:
> [..]
> > +		/*
> > +		 * This function could be called in swapout/fs write path
> > +		 * so we couldn't use GFP_FS|IO. And it assumes we already
> > +		 * have at least one stream in zram initialization so we
> > +		 * don't do best effort to allocate more stream in here.
> > +		 * A default stream will work well without further multiple
> > +		 * stream. That's why we use __GFP_NORETRY|NOWARN|NOMEMALLOC.
> > +		 */
> > +		zstrm = zcomp_strm_alloc(comp, GFP_NOIO|__GFP_NORETRY|
> > +					__GFP_NOWARN|__GFP_NOMEMALLOC);
> [..]
> 
> I think that applying 3/3 before 2/3 will be a simpler (and probably a better)
> thing to do. We fitst extend zcomp interface and pass flags (without any
> functional change) and then extend the flags and introduce vmalloc fallback.

The reason I ordered such way is that I wanted to discuss [2/3] as
stable material after I get your ACK. It solves real problem in android platform
which is real fact and I think it's enough small to send stable tree.
What do you think?

> 
> So we don't have to add comments to lz4/lzo backend that are getting (re-)moved
> in the very next commit.

Fair enough if you don't agree sending [2/3] to stable.

> 
> I will send swapped 2 and 3 patches shortly (I didn't change commit
> messages and SoBs). Please take a look.

Thanks!

> 
> > -	/*
> > -	 * This function could be called in swapout/fs write path
> > -	 * so we couldn't use GFP_FS|IO. And it assumes we already
> > -	 * have at least one stream in zram initialization so we
> > -	 * don't do best effort to allocate more stream in here.
> > -	 * A default stream will work well without further multiple
> > -	 * stream. That's why we use  __GFP_NORETRY|NOWARN|NOMEMALLOC.
> > -	 */
> > -	ret = kzalloc(LZ4_MEM_COMPRESS, GFP_NOIO|__GFP_NORETRY|
> > -					__GFP_NOWARN|__GFP_NOMEMALLOC);
> > +	ret = kmalloc(LZ4_MEM_COMPRESS, flags);
> >  	if (!ret)
> > -		ret = __vmalloc(LZ4_MEM_COMPRESS, GFP_NOIO|__GFP_NORETRY|
> > -						__GFP_NOWARN|__GFP_NOMEMALLOC|
> > -						__GFP_ZERO,
> > -						PAGE_KERNEL);
> > +		ret = __vmalloc(LZ4_MEM_COMPRESS, flags, PAGE_KERNEL);
> [..]
> 
> __vmalloc() is still missing __GFP_HIGHMEM

Argh, Sorry.