On Wed, Aug 12, 2020 at 10:54:08AM -0700, Linus Torvalds wrote: > On Tue, Aug 11, 2020 at 6:35 PM Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx> wrote: > > > > From: Nick Terrell <terrelln@xxxxxx> > > Subject: lz4: fix kernel decompression speed > > > > This patch replaces all memcpy() calls with LZ4_memcpy() which calls > > __builtin_memcpy() so the compiler can inline it. > > Wasn't this LZ4_memcpy() wrapper made unnecessary by just making > memcpy() work properly on its own? > > So I'm dropping this patch. > > If it turns out that I mis-remembered (or mis-understood), please re-send. Nick? > > Linus I assume you're referring to my patch [0]. It hasn't been picked up yet, and it is only for x86. I think Nick wanted his patch to be merged as well [1], because it will also address other architectures (eg it looks like at least s390 LZ4 probably has the same issue), and is already in upstream LZ4. Thanks. [0] https://lore.kernel.org/lkml/20200804234817.3922187-1-nivedita@xxxxxxxxxxxx/ [1] https://lore.kernel.org/lkml/3961E1BD-8F58-4240-A3B3-B7032A405B42@xxxxxx/
