Re: [PATCH v3] zram: try vmalloc() after kmalloc()

[Minchan Kim <minchan@kernel.org>]

Hello Andrew,

On Mon, Nov 23, 2015 at 02:52:26PM -0800, Andrew Morton wrote:
> On Mon, 23 Nov 2015 15:21:15 +0900 Kyeongdon Kim <kyeongdon.kim@lge.com> wrote:
> 
> > When we're using LZ4 multi compression streams for zram swap,
> > we found out page allocation failure message in system running test.
> > That was not only once, but a few(2 - 5 times per test).
> > Also, some failure cases were continually occurring to try allocation
> > order 3.
> > 
> > In order to make parallel compression private data, we should call
> > kzalloc() with order 2/3 in runtime(lzo/lz4). But if there is no order
> >  2/3 size memory to allocate in that time, page allocation fails.
> > This patch makes to use vmalloc() as fallback of kmalloc(), this
> > prevents page alloc failure warning.
> > 
> > After using this, we never found warning message in running test, also
> > It could reduce process startup latency about 60-120ms in each case.
> > 
> > ...
> >
> > --- a/drivers/block/zram/zcomp_lz4.c
> > +++ b/drivers/block/zram/zcomp_lz4.c
> > @@ -10,17 +10,25 @@
> >  #include <linux/kernel.h>
> >  #include <linux/slab.h>
> >  #include <linux/lz4.h>
> > +#include <linux/vmalloc.h>
> > +#include <linux/mm.h>
> >  
> >  #include "zcomp_lz4.h"
> >  
> >  static void *zcomp_lz4_create(void)
> >  {
> > -	return kzalloc(LZ4_MEM_COMPRESS, GFP_KERNEL);
> > +	void *ret;
> > +
> > +	ret = kzalloc(LZ4_MEM_COMPRESS,
> > +			__GFP_NORETRY|__GFP_NOWARN|__GFP_NOMEMALLOC);
> > +	if (!ret)
> > +		ret = vzalloc(LZ4_MEM_COMPRESS);
> > +	return ret;
> >  }
> 
> What's the reasoning behind the modification to the gfp flags?
> 
> It clears __GFP_FS, __GFP_IO and even __GFP_WAIT.  I suspect the latter
> two (at least) can be retained.  And given that vmalloc() uses

This function is used in swapout and fs write path so we couldn't use
those flags.

> GFP_KERNEL, what's the point in clearing those flags for the kmalloc()
> case?

When I reviewed this patch, I wanted to fix it with another patch
because we should handle another places in zcomp and Sergey sent it
today. But I think Sergey's patch is stable material so I hope
Kyeongdon resend this patch with fixing vmalloc part.

> 
> If this change (or something like it) remains in place, it should have
> a comment which fully explains the reasons, please.

Kyeongdon, Could you resend this patch with fixing vzalloc part and
adding comment? 

> 

-- 
Kind regards,
Minchan Kim