From: Glauber Costa <gcosta@redhat.com>
To: linux-kernel@vger.kernel.org
Cc: akpm@linux-foundation.org, glommer@gmail.com, mingo@elte.hu,
tglx@linutronix.de, kvm-devel@lists.sourceforge.net,
amit.shah@qumranet.com, avi@qumranet.com,
Glauber Costa <gcosta@redhat.com>
Subject: [PATCH 21/28] x86: retry allocation if failed
Date: Tue, 8 Apr 2008 13:21:03 -0300 [thread overview]
Message-ID: <12076718073024-git-send-email-gcosta@redhat.com> (raw)
In-Reply-To: <12076716702129-git-send-email-gcosta@redhat.com>
This patch puts in the code to retry allocation in case it fails. By its
own, it does not make much sense but making the code look like x86_64.
But later patches in this series will make we try to allocate from
zones other than DMA first, which will possibly fail.
Signed-off-by: Glauber Costa <gcosta@redhat.com>
---
arch/x86/kernel/pci-dma_32.c | 34 +++++++++++++++++++++++++++++-----
1 files changed, 29 insertions(+), 5 deletions(-)
diff --git a/arch/x86/kernel/pci-dma_32.c b/arch/x86/kernel/pci-dma_32.c
index 0d630ae..f6cf434 100644
--- a/arch/x86/kernel/pci-dma_32.c
+++ b/arch/x86/kernel/pci-dma_32.c
@@ -66,6 +66,8 @@ void *dma_alloc_coherent(struct device *dev, size_t size,
struct page *page;
dma_addr_t bus;
int order = get_order(size);
+ unsigned long dma_mask = 0;
+
/* ignore region specifiers */
gfp &= ~(__GFP_DMA | __GFP_HIGHMEM);
@@ -75,15 +77,37 @@ void *dma_alloc_coherent(struct device *dev, size_t size,
if (dev == NULL || (dev->coherent_dma_mask < 0xffffffff))
gfp |= GFP_DMA;
+ dma_mask = dev->coherent_dma_mask;
+ if (dma_mask == 0)
+ dma_mask = DMA_32BIT_MASK;
+
+ again:
page = dma_alloc_pages(dev, gfp, order);
if (page == NULL)
return NULL;
- ret = page_address(page);
- bus = page_to_phys(page);
-
- memset(ret, 0, size);
- *dma_handle = bus;
+ {
+ int high, mmu;
+ bus = page_to_phys(page);
+ ret = page_address(page);
+ high = (bus + size) >= dma_mask;
+ mmu = high;
+ if (force_iommu && !(gfp & GFP_DMA))
+ mmu = 1;
+ else if (high) {
+ free_pages((unsigned long)ret,
+ get_order(size));
+
+ /* Don't use the 16MB ZONE_DMA unless absolutely
+ needed. It's better to use remapping first. */
+ if (dma_mask < DMA_32BIT_MASK && !(gfp & GFP_DMA)) {
+ gfp = (gfp & ~GFP_DMA32) | GFP_DMA;
+ goto again;
+ }
+ }
+ memset(ret, 0, size);
+ *dma_handle = bus;
+ }
return ret;
}
--
1.5.0.6
next prev parent reply other threads:[~2008-04-08 16:21 UTC|newest]
Thread overview: 29+ messages / expand[flat|nested] mbox.gz Atom feed top
2008-04-08 16:20 [PATCH 0/28] integrate dma_ops Glauber Costa
2008-04-08 16:20 ` [PATCH 01/28] x86: introduce pci-dma.c Glauber Costa
2008-04-08 16:20 ` [PATCH 02/28] x86: delete empty functions from pci-nommu_64.c Glauber Costa
2008-04-08 16:20 ` [PATCH 03/28] x86: implement mapping_error in pci-nommu_64.c Glauber Costa
2008-04-08 16:20 ` [PATCH 04/28] x86: Add flush_write_buffers in nommu functions Glauber Costa
2008-04-08 16:20 ` [PATCH 05/28] x86: use sg_phys in x86_64 Glauber Costa
2008-04-08 16:20 ` [PATCH 06/28] x86: use dma_length in i386 Glauber Costa
2008-04-08 16:20 ` [PATCH 07/28] x86: use WARN_ON in mapping functions Glauber Costa
2008-04-08 16:20 ` [PATCH 08/28] x86: move definition to pci-dma.c Glauber Costa
2008-04-08 16:20 ` [PATCH 09/28] x86: move initialization functions " Glauber Costa
2008-04-08 16:20 ` [PATCH 10/28] x86: unify pci-nommu Glauber Costa
2008-04-08 16:20 ` [PATCH 11/28] x86: move pci fixup to pci-dma.c Glauber Costa
2008-04-08 16:20 ` [PATCH 12/28] x86: move x86_64-specific to common code Glauber Costa
2008-04-08 16:20 ` [PATCH 13/28] x86: merge dma_supported Glauber Costa
2008-04-08 16:20 ` [PATCH 14/28] x86: merge iommu initialization parameters Glauber Costa
2008-04-08 16:20 ` [PATCH 15/28] x86: move dma_coherent functions to pci-dma.c Glauber Costa
2008-04-08 16:20 ` [PATCH 16/28] x86: isolate coherent mapping functions Glauber Costa
2008-04-08 16:20 ` [PATCH 17/28] x86: adjust dma_free_coherent for i386 Glauber Costa
2008-04-08 16:21 ` [PATCH 18/28] x86: move bad_dma_address Glauber Costa
2008-04-08 16:21 ` [PATCH 19/28] x86: remove virt_to_bus in pci-dma_64.c Glauber Costa
2008-04-08 16:21 ` [PATCH 20/28] x86: use numa allocation function in i386 Glauber Costa
2008-04-08 16:21 ` Glauber Costa [this message]
2008-04-08 16:21 ` [PATCH 22/28] x86: use a fallback dev for i386 Glauber Costa
2008-04-08 16:21 ` [PATCH 23/28] x86: don't try to allocate from DMA zone at first Glauber Costa
2008-04-08 16:21 ` [PATCH 24/28] x86: unify gfp masks Glauber Costa
2008-04-08 16:21 ` [PATCH 25/28] x86: remove kludge from x86_64 Glauber Costa
2008-04-08 16:21 ` [PATCH 26/28] x86: return conditional to mmu Glauber Costa
2008-04-08 16:21 ` [PATCH 27/28] x86: don't do dma if mask is NULL Glauber Costa
2008-04-08 16:21 ` [PATCH 28/28] x86: integrate pci-dma.c Glauber Costa
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=12076718073024-git-send-email-gcosta@redhat.com \
--to=gcosta@redhat.com \
--cc=akpm@linux-foundation.org \
--cc=amit.shah@qumranet.com \
--cc=avi@qumranet.com \
--cc=glommer@gmail.com \
--cc=kvm-devel@lists.sourceforge.net \
--cc=linux-kernel@vger.kernel.org \
--cc=mingo@elte.hu \
--cc=tglx@linutronix.de \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox