[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20170209110211.GB3575@bierbaron.springfield.local>
Date: Thu, 9 Feb 2017 12:02:11 +0100
From: Sven Schmidt <4sschmid@...ormatik.uni-hamburg.de>
To: Eric Biggers <ebiggers3@...il.com>
Cc: Minchan Kim <minchan@...nel.org>, akpm@...ux-foundation.org,
bongkyu.kim@....com, rsalvaterra@...il.com,
sergey.senozhatsky@...il.com, gregkh@...uxfoundation.org,
linux-kernel@...r.kernel.org, herbert@...dor.apana.org.au,
davem@...emloft.net, linux-crypto@...r.kernel.org,
anton@...msg.org, ccross@...roid.com, keescook@...omium.org,
tony.luck@...el.com
Subject: Re: [PATCH v7 0/5] Update LZ4 compressor module
Hey Eric,
On Wed, Feb 08, 2017 at 04:24:36PM -0800, Eric Biggers wrote:
> On Thu, Feb 09, 2017 at 08:31:21AM +0900, Minchan Kim wrote:
> >
> > Today, I did zram-lz4 performance test with fio in current mmotm and
> > found it makes regression about 20%.
> >
>
> This may or may not be the cause of the specific regression you're observing,
> but I just noticed that the proposed patch drops a lot of FORCEINLINE
> annotations from upstream LZ4. The FORCEINLINE's are there for a reason,
> especially for the main decompression and compression functions which are
> basically "templates" that take in different sets of constant parameters, and
> should be left in. We should #define FORCEINLINE to __always_inline somewhere,
> or just do a s/FORCEINLINE/__always_inline/g.
>
I generally just replaced "FORCE_INLINE" by "static inline". At least I thought so.
I rechecked and realised, I missed at least two of them (why did I not just use "search+replace"?).
So I think it's maybe safer and easier to eventually just use "FORCE_INLINE"
with the definition you suggested. Will try that.
> Note that the upstream LZ4 code is very carefully optimized, so we should not,
> in general, be changing things like when functions are force-inlined, what the
> hash table size is, etc.
>
> [Also, for some reason linux-crypto is apparently still not receiving patch 1/5
> in the series. It's missing from the linux-crypto archive at
> http://www.spinics.net/lists/linux-crypto/, so it's not just me.]
>
I don't really know what to do about this. I think the matter is the size of the E-Mail.
Are there filters or something like that? Since in linux-kernel the patch seems to get delivered.
I could otherwise CC you if you wish.
Thanks,
Sven
Powered by blists - more mailing lists