[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20240823080736.GA3068@lst.de>
Date: Fri, 23 Aug 2024 10:07:36 +0200
From: Christoph Hellwig <hch@....de>
To: Christian Lamparter <chunkeey@...il.com>
Cc: Christoph Hellwig <hch@....de>,
LEROY Christophe <christophe.leroy2@...soprasteria.com>,
Benjamin Herrenschmidt <benh@...nel.crashing.org>,
Paul Mackerras <paulus@...ba.org>,
Michael Ellerman <mpe@...erman.id.au>,
"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
"linuxppc-dev@...ts.ozlabs.org" <linuxppc-dev@...ts.ozlabs.org>,
Stan Johnson <userm57@...oo.com>,
Finn Thain <fthain@...ux-m68k.org>, Vinod Koul <vkoul@...nel.org>,
dmaengine@...r.kernel.org
Subject: Re: [PATCH v2] powerpc: warn on emulation of dcbz instruction in
kernel mode
dma_alloc_from_pool is the allocator used when the caller can't
sleep, and as that is reusing memory it really has to call memset
or a memset-like function on the already uncached memory
unfortunately. The dma engine operation that is doing this allocation
is documented as not being able to sleep, which is a bit unfortunate
as the storage driver above it could sleep just fine.
Adding the dmaengine maintainer and list if there is a way to pass a gfp
flag or some other indicator that the implementations can sleep,
which would avoid the need to use the not very scalable dma pool,
and thus also the need to memset on uncached memory.
Powered by blists - more mailing lists