From patchwork Tue Jun 21 09:26:27 2011 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Per Forlin X-Patchwork-Id: 900932 Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by demeter1.kernel.org (8.14.4/8.14.4) with ESMTP id p5L9QVHX024574 for ; Tue, 21 Jun 2011 09:26:31 GMT Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1755747Ab1FUJ03 (ORCPT ); Tue, 21 Jun 2011 05:26:29 -0400 Received: from mail-qw0-f46.google.com ([209.85.216.46]:56373 "EHLO mail-qw0-f46.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1755155Ab1FUJ03 (ORCPT ); Tue, 21 Jun 2011 05:26:29 -0400 Received: by qwk3 with SMTP id 3so1180435qwk.19 for ; Tue, 21 Jun 2011 02:26:28 -0700 (PDT) MIME-Version: 1.0 Received: by 10.224.220.82 with SMTP id hx18mr4746872qab.320.1308648387888; Tue, 21 Jun 2011 02:26:27 -0700 (PDT) Received: by 10.224.2.193 with HTTP; Tue, 21 Jun 2011 02:26:27 -0700 (PDT) In-Reply-To: References: <1308518257-9783-1-git-send-email-per.forlin@linaro.org> <20110621075319.GN26089@n2100.arm.linux.org.uk> Date: Tue, 21 Jun 2011 11:26:27 +0200 Message-ID: Subject: Re: [PATCH v6 00/11] mmc: use nonblock mmc requests to minimize latency From: Per Forlin To: Russell King - ARM Linux Cc: linux-mmc@vger.kernel.org, linux-arm-kernel@lists.infradead.org, Nicolas Pitre , linaro-dev@lists.linaro.org, Venkatraman S , Chris Ball Sender: linux-mmc-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-mmc@vger.kernel.org X-Greylist: IP, sender and recipient auto-whitelisted, not delayed by milter-greylist-4.2.6 (demeter1.kernel.org [140.211.167.41]); Tue, 21 Jun 2011 09:26:31 +0000 (UTC) On 21 June 2011 10:09, Per Forlin wrote: > On 21 June 2011 09:53, Russell King - ARM Linux wrote: >> On Sun, Jun 19, 2011 at 11:17:26PM +0200, Per Forlin wrote: >>> How significant is the cache maintenance over head? >> >> Per, >> >> Can you measure how much difference this has before and after your >> patch set please? > Absolutely, I can run the mmc_tests to get the measurement. The cache > affect is greater the faster the flash memory is. Currently I only > have access to a SD card (20 MiB/S). By the end of this week I can run > on eMMC (45 MiB/s) if this will be needed. > Russel, Here are the results. mmc_test results without your DSB patch: mmc0: Starting tests of card mmc0:80ca... mmc0: Test case 37. Write performance with blocking req 4k to 4MB... mmc0: Transfer of 32768 x 8 sectors (32768 x 4 KiB) took 17.907140069 seconds (7495 kB/s, 7319 KiB/s, 1829.88 IOPS) mmc0: Transfer of 16384 x 16 sectors (16384 x 8 KiB) took 10.977203519 seconds (12226 kB/s, 11940 KiB/s, 1492.54 IOPS) mmc0: Transfer of 8192 x 32 sectors (8192 x 16 KiB) took 8.618723194 seconds (15572 kB/s, 15207 KiB/s, 950.48 IOPS) mmc0: Transfer of 4096 x 64 sectors (4096 x 32 KiB) took 7.452392708 seconds (18010 kB/s, 17587 KiB/s, 549.62 IOPS) mmc0: Transfer of 2048 x 128 sectors (2048 x 64 KiB) took 6.839447152 seconds (19624 kB/s, 19164 KiB/s, 299.43 IOPS) mmc0: Transfer of 1024 x 256 sectors (1024 x 128 KiB) took 6.533447450 seconds (20543 kB/s, 20061 KiB/s, 156.73 IOPS) mmc0: Transfer of 512 x 512 sectors (512 x 256 KiB) took 6.355529943 seconds (21118 kB/s, 20623 KiB/s, 80.55 IOPS) mmc0: Transfer of 256 x 1024 sectors (256 x 512 KiB) took 6.227417019 seconds (21552 kB/s, 21047 KiB/s, 41.10 IOPS) mmc0: Transfer of 128 x 2048 sectors (128 x 1024 KiB) took 6.047821091 seconds (22192 kB/s, 21672 KiB/s, 21.16 IOPS) mmc0: Transfer of 32 x 8192 sectors (32 x 4096 KiB) took 5.983120236 seconds (22432 kB/s, 21906 KiB/s, 5.34 IOPS) mmc0: Result: OK mmc0: Tests completed. mmc0: Starting tests of card mmc0:80ca... mmc0: Test case 38. Write performance with non-blocking req 4k to 4MB... mmc0: Transfer of 32768 x 8 sectors (32768 x 4 KiB) took 17.004930158 seconds (7892 kB/s, 7707 KiB/s, 1926.97 IOPS) mmc0: Transfer of 16384 x 16 sectors (16384 x 8 KiB) took 10.397338972 seconds (12908 kB/s, 12606 KiB/s, 1575.78 IOPS) mmc0: Transfer of 8192 x 32 sectors (8192 x 16 KiB) took 8.127319360 seconds (16514 kB/s, 16127 KiB/s, 1007.95 IOPS) mmc0: Transfer of 4096 x 64 sectors (4096 x 32 KiB) took 7.061096329 seconds (19008 kB/s, 18562 KiB/s, 580.07 IOPS) mmc0: Transfer of 2048 x 128 sectors (2048 x 64 KiB) took 6.503535845 seconds (20637 kB/s, 20153 KiB/s, 314.90 IOPS) mmc0: Transfer of 1024 x 256 sectors (1024 x 128 KiB) took 6.222897631 seconds (21568 kB/s, 21062 KiB/s, 164.55 IOPS) mmc0: Transfer of 512 x 512 sectors (512 x 256 KiB) took 6.082733285 seconds (22065 kB/s, 21548 KiB/s, 84.17 IOPS) mmc0: Transfer of 256 x 1024 sectors (256 x 512 KiB) took 5.928009056 seconds (22641 kB/s, 22110 KiB/s, 43.18 IOPS) mmc0: Transfer of 128 x 2048 sectors (128 x 1024 KiB) took 5.891113751 seconds (22783 kB/s, 22249 KiB/s, 21.72 IOPS) mmc0: Transfer of 32 x 8192 sectors (32 x 4096 KiB) took 5.878531233 seconds (22831 kB/s, 22296 KiB/s, 5.44 IOPS) mmc0: Result: OK mmc0: Tests completed. mmc0: Starting tests of card mmc0:80ca... mmc0: Test case 39. Read performance with blocking req 4k to 4MB... mmc0: Transfer of 32768 x 8 sectors (32768 x 4 KiB) took 20.904750140 seconds (6420 kB/s, 6269 KiB/s, 1567.49 IOPS) mmc0: Transfer of 16384 x 16 sectors (16384 x 8 KiB) took 12.929870605 seconds (10380 kB/s, 10137 KiB/s, 1267.14 IOPS) mmc0: Transfer of 8192 x 32 sectors (8192 x 16 KiB) took 10.115753174 seconds (13268 kB/s, 12957 KiB/s, 809.82 IOPS) mmc0: Transfer of 4096 x 64 sectors (4096 x 32 KiB) took 7.533538819 seconds (17816 kB/s, 17398 KiB/s, 543.70 IOPS) mmc0: Transfer of 2048 x 128 sectors (2048 x 64 KiB) took 6.937011718 seconds (19348 kB/s, 18894 KiB/s, 295.22 IOPS) mmc0: Transfer of 1024 x 256 sectors (1024 x 128 KiB) took 6.638824464 seconds (20217 kB/s, 19743 KiB/s, 154.24 IOPS) mmc0: Transfer of 512 x 512 sectors (512 x 256 KiB) took 6.489288330 seconds (20682 kB/s, 20198 KiB/s, 78.89 IOPS) mmc0: Transfer of 256 x 1024 sectors (256 x 512 KiB) took 6.414489746 seconds (20924 kB/s, 20433 KiB/s, 39.90 IOPS) mmc0: Transfer of 128 x 2048 sectors (128 x 1024 KiB) took 6.376800426 seconds (21047 kB/s, 20554 KiB/s, 20.07 IOPS) mmc0: Transfer of 32 x 8192 sectors (32 x 4096 KiB) took 6.348991821 seconds (21140 kB/s, 20644 KiB/s, 5.04 IOPS) mmc0: Result: OK mmc0: Tests completed. mmc0: Starting tests of card mmc0:80ca... mmc0: Test case 40. Read performance with non-blocking req 4k to 4MB... mmc0: Transfer of 32768 x 8 sectors (32768 x 4 KiB) took 20.906376527 seconds (6419 kB/s, 6269 KiB/s, 1567.36 IOPS) mmc0: Transfer of 16384 x 16 sectors (16384 x 8 KiB) took 12.929779053 seconds (10380 kB/s, 10137 KiB/s, 1267.15 IOPS) mmc0: Transfer of 8192 x 32 sectors (8192 x 16 KiB) took 10.119873047 seconds (13262 kB/s, 12951 KiB/s, 809.49 IOPS) mmc0: Transfer of 4096 x 64 sectors (4096 x 32 KiB) took 7.501770019 seconds (17891 kB/s, 17472 KiB/s, 546.00 IOPS) mmc0: Transfer of 2048 x 128 sectors (2048 x 64 KiB) took 6.797882080 seconds (19744 kB/s, 19281 KiB/s, 301.27 IOPS) mmc0: Transfer of 1024 x 256 sectors (1024 x 128 KiB) took 6.293121338 seconds (21327 kB/s, 20827 KiB/s, 162.71 IOPS) mmc0: Transfer of 512 x 512 sectors (512 x 256 KiB) took 5.952606200 seconds (22547 kB/s, 22019 KiB/s, 86.01 IOPS) mmc0: Transfer of 256 x 1024 sectors (256 x 512 KiB) took 5.862152101 seconds (22895 kB/s, 22359 KiB/s, 43.66 IOPS) mmc0: Transfer of 128 x 2048 sectors (128 x 1024 KiB) took 5.818847175 seconds (23066 kB/s, 22525 KiB/s, 21.99 IOPS) mmc0: Transfer of 32 x 8192 sectors (32 x 4096 KiB) took 5.798218390 seconds (23148 kB/s, 22605 KiB/s, 5.51 IOPS) mmc0: Result: OK mmc0: Tests completed. mmc_test results with your DSB patch: mmc0: Starting tests of card mmc0:80ca... mmc0: Test case 37. Write performance with blocking req 4k to 4MB... mmc0: Transfer of 32768 x 8 sectors (32768 x 4 KiB) took 17.912285550 seconds (7493 kB/s, 7317 KiB/s, 1829.35 IOPS) mmc0: Transfer of 16384 x 16 sectors (16384 x 8 KiB) took 10.992614823 seconds (12209 kB/s, 11923 KiB/s, 1490.45 IOPS) mmc0: Transfer of 8192 x 32 sectors (8192 x 16 KiB) took 8.670936194 seconds (15479 kB/s, 15116 KiB/s, 944.76 IOPS) mmc0: Transfer of 4096 x 64 sectors (4096 x 32 KiB) took 7.448752639 seconds (18018 kB/s, 17596 KiB/s, 549.89 IOPS) mmc0: Transfer of 2048 x 128 sectors (2048 x 64 KiB) took 6.837432905 seconds (19629 kB/s, 19169 KiB/s, 299.52 IOPS) mmc0: Transfer of 1024 x 256 sectors (1024 x 128 KiB) took 6.510650765 seconds (20615 kB/s, 20131 KiB/s, 157.28 IOPS) mmc0: Transfer of 512 x 512 sectors (512 x 256 KiB) took 6.343047841 seconds (21159 kB/s, 20663 KiB/s, 80.71 IOPS) mmc0: Transfer of 256 x 1024 sectors (256 x 512 KiB) took 6.275632327 seconds (21387 kB/s, 20885 KiB/s, 40.79 IOPS) mmc0: Transfer of 128 x 2048 sectors (128 x 1024 KiB) took 6.051895663 seconds (22177 kB/s, 21658 KiB/s, 21.15 IOPS) mmc0: Transfer of 32 x 8192 sectors (32 x 4096 KiB) took 5.992395203 seconds (22398 kB/s, 21873 KiB/s, 5.34 IOPS) mmc0: Result: OK mmc0: Tests completed. mmc0: Starting tests of card mmc0:80ca... mmc0: Test case 38. Write performance with non-blocking req 4k to 4MB... mmc0: Transfer of 32768 x 8 sectors (32768 x 4 KiB) took 17.019586188 seconds (7886 kB/s, 7701 KiB/s, 1925.31 IOPS) mmc0: Transfer of 16384 x 16 sectors (16384 x 8 KiB) took 10.377655096 seconds (12933 kB/s, 12630 KiB/s, 1578.77 IOPS) mmc0: Transfer of 8192 x 32 sectors (8192 x 16 KiB) took 8.172790531 seconds (16422 kB/s, 16037 KiB/s, 1002.35 IOPS) mmc0: Transfer of 4096 x 64 sectors (4096 x 32 KiB) took 7.069458097 seconds (18985 kB/s, 18540 KiB/s, 579.39 IOPS) mmc0: Transfer of 2048 x 128 sectors (2048 x 64 KiB) took 6.498779387 seconds (20652 kB/s, 20168 KiB/s, 315.13 IOPS) mmc0: Transfer of 1024 x 256 sectors (1024 x 128 KiB) took 6.220800166 seconds (21575 kB/s, 21069 KiB/s, 164.60 IOPS) mmc0: Transfer of 512 x 512 sectors (512 x 256 KiB) took 6.040708413 seconds (22218 kB/s, 21698 KiB/s, 84.75 IOPS) mmc0: Transfer of 256 x 1024 sectors (256 x 512 KiB) took 5.946899457 seconds (22569 kB/s, 22040 KiB/s, 43.04 IOPS) mmc0: Transfer of 128 x 2048 sectors (128 x 1024 KiB) took 5.927886710 seconds (22641 kB/s, 22111 KiB/s, 21.59 IOPS) mmc0: Transfer of 32 x 8192 sectors (32 x 4096 KiB) took 5.878386087 seconds (22832 kB/s, 22297 KiB/s, 5.44 IOPS) mmc0: Result: OK mmc0: Tests completed. mmc0: Starting tests of card mmc0:80ca... mmc0: Test case 39. Read performance with blocking req 4k to 4MB... mmc0: Transfer of 32768 x 8 sectors (32768 x 4 KiB) took 20.829314216 seconds (6443 kB/s, 6292 KiB/s, 1573.16 IOPS) mmc0: Transfer of 16384 x 16 sectors (16384 x 8 KiB) took 12.875244140 seconds (10424 kB/s, 10180 KiB/s, 1272.51 IOPS) mmc0: Transfer of 8192 x 32 sectors (8192 x 16 KiB) took 10.073059082 seconds (13324 kB/s, 13012 KiB/s, 813.25 IOPS) mmc0: Transfer of 4096 x 64 sectors (4096 x 32 KiB) took 7.550659181 seconds (17775 kB/s, 17359 KiB/s, 542.46 IOPS) mmc0: Transfer of 2048 x 128 sectors (2048 x 64 KiB) took 6.942535401 seconds (19332 kB/s, 18879 KiB/s, 294.99 IOPS) mmc0: Transfer of 1024 x 256 sectors (1024 x 128 KiB) took 6.645233154 seconds (20197 kB/s, 19724 KiB/s, 154.09 IOPS) mmc0: Transfer of 512 x 512 sectors (512 x 256 KiB) took 6.495941164 seconds (20661 kB/s, 20177 KiB/s, 78.81 IOPS) mmc0: Transfer of 256 x 1024 sectors (256 x 512 KiB) took 6.421081542 seconds (20902 kB/s, 20412 KiB/s, 39.86 IOPS) mmc0: Transfer of 128 x 2048 sectors (128 x 1024 KiB) took 6.383514604 seconds (21025 kB/s, 20532 KiB/s, 20.05 IOPS) mmc0: Transfer of 32 x 8192 sectors (32 x 4096 KiB) took 6.355718936 seconds (21117 kB/s, 20622 KiB/s, 5.03 IOPS) mmc0: Result: OK mmc0: Tests completed. mmc0: Starting tests of card mmc0:80ca... mmc0: Test case 40. Read performance with non-blocking req 4k to 4MB... mmc0: Transfer of 32768 x 8 sectors (32768 x 4 KiB) took 20.832669187 seconds (6442 kB/s, 6291 KiB/s, 1572.91 IOPS) mmc0: Transfer of 16384 x 16 sectors (16384 x 8 KiB) took 12.884582520 seconds (10416 kB/s, 10172 KiB/s, 1271.59 IOPS) mmc0: Transfer of 8192 x 32 sectors (8192 x 16 KiB) took 10.076812745 seconds (13319 kB/s, 13007 KiB/s, 812.95 IOPS) mmc0: Transfer of 4096 x 64 sectors (4096 x 32 KiB) took 7.471252441 seconds (17964 kB/s, 17543 KiB/s, 548.23 IOPS) mmc0: Transfer of 2048 x 128 sectors (2048 x 64 KiB) took 6.765075684 seconds (19839 kB/s, 19374 KiB/s, 302.73 IOPS) mmc0: Transfer of 1024 x 256 sectors (1024 x 128 KiB) took 6.259826661 seconds (21441 kB/s, 20938 KiB/s, 163.58 IOPS) mmc0: Transfer of 512 x 512 sectors (512 x 256 KiB) took 5.948974608 seconds (22561 kB/s, 22032 KiB/s, 86.06 IOPS) mmc0: Transfer of 256 x 1024 sectors (256 x 512 KiB) took 5.860260010 seconds (22903 kB/s, 22366 KiB/s, 43.68 IOPS) mmc0: Transfer of 128 x 2048 sectors (128 x 1024 KiB) took 5.817993397 seconds (23069 kB/s, 22528 KiB/s, 22.00 IOPS) mmc0: Transfer of 32 x 8192 sectors (32 x 4096 KiB) took 5.798185906 seconds (23148 kB/s, 22605 KiB/s, 5.51 IOPS) mmc0: Result: OK mmc0: Tests completed. In case I did any mistakes applying your patch manually. Here is your dsb patch on top of 3.0-rc4. } @@ -572,6 +573,7 @@ int dma_map_sg(struct device *dev, struct scatterlist *sg, int nents, if (dma_mapping_error(dev, s->dma_address)) goto bad_mapping; } + __dma_sync(); debug_dma_map_sg(dev, sg, nents, nents, dir); return nents; @@ -602,6 +604,7 @@ void dma_unmap_sg(struct device *dev, struct scatterlist *sg, int nents, for_each_sg(sg, s, nents, i) __dma_unmap_page(dev, sg_dma_address(s), sg_dma_len(s), dir); + __dma_sync(); } EXPORT_SYMBOL(dma_unmap_sg); @@ -626,6 +629,7 @@ void dma_sync_sg_for_cpu(struct device *dev, struct scatterlist *sg, __dma_page_dev_to_cpu(sg_page(s), s->offset, s->length, dir); } + __dma_sync(); debug_dma_sync_sg_for_cpu(dev, sg, nents, dir); } @@ -652,6 +656,7 @@ void dma_sync_sg_for_device(struct device *dev, struct scatterlist *sg, __dma_page_cpu_to_dev(sg_page(s), s->offset, s->length, dir); } + __dma_sync(); debug_dma_sync_sg_for_device(dev, sg, nents, dir); } diff --git a/arch/arm/include/asm/dma-mapping.h b/arch/arm/include/asm/dma-mapping.h index 4fff837..ad14c2b 100644 --- a/arch/arm/include/asm/dma-mapping.h +++ b/arch/arm/include/asm/dma-mapping.h @@ -115,6 +115,11 @@ static inline void __dma_page_dev_to_cpu(struct page *page, unsigned long off, ___dma_page_dev_to_cpu(page, off, size, dir); } +static inline void __dma_sync(void) +{ + dsb(); +} + /* * Return whether the given device DMA address mask can be supported * properly. For example, if your device can only drive the low 24-bits @@ -378,6 +383,7 @@ static inline dma_addr_t dma_map_single(struct device *dev, void *cpu_addr, BUG_ON(!valid_dma_direction(dir)); addr = __dma_map_single(dev, cpu_addr, size, dir); + __dma_sync(); debug_dma_map_page(dev, virt_to_page(cpu_addr), (unsigned long)cpu_addr & ~PAGE_MASK, size, dir, addr, true); @@ -407,6 +413,7 @@ static inline dma_addr_t dma_map_page(struct device *dev, struct page *page, BUG_ON(!valid_dma_direction(dir)); addr = __dma_map_page(dev, page, offset, size, dir); + __dma_sync(); debug_dma_map_page(dev, page, offset, size, dir, addr, false); return addr; @@ -431,6 +438,7 @@ static inline void dma_unmap_single(struct device *dev, dma_addr_t handle, { debug_dma_unmap_page(dev, handle, size, dir, true); __dma_unmap_single(dev, handle, size, dir); + __dma_sync(); } /** @@ -452,6 +460,7 @@ static inline void dma_unmap_page(struct device *dev, dma_addr_t handle, { debug_dma_unmap_page(dev, handle, size, dir, false); __dma_unmap_page(dev, handle, size, dir); + __dma_sync(); } /** @@ -498,6 +507,7 @@ static inline void dma_sync_single_range_for_device(struct device *dev, return; __dma_single_cpu_to_dev(dma_to_virt(dev, handle) + offset, size, dir); + __dma_sync(); } static inline void dma_sync_single_for_cpu(struct device *dev, diff --git a/arch/arm/mm/cache-fa.S b/arch/arm/mm/cache-fa.S index 1fa6f71..6eeb734 100644 --- a/arch/arm/mm/cache-fa.S +++ b/arch/arm/mm/cache-fa.S @@ -179,8 +179,6 @@ fa_dma_inv_range: add r0, r0, #CACHE_DLINESIZE cmp r0, r1 blo 1b - mov r0, #0 - mcr p15, 0, r0, c7, c10, 4 @ drain write buffer mov pc, lr /* @@ -197,8 +195,6 @@ fa_dma_clean_range: add r0, r0, #CACHE_DLINESIZE cmp r0, r1 blo 1b - mov r0, #0 - mcr p15, 0, r0, c7, c10, 4 @ drain write buffer mov pc, lr /* @@ -212,8 +208,6 @@ ENTRY(fa_dma_flush_range) add r0, r0, #CACHE_DLINESIZE cmp r0, r1 blo 1b - mov r0, #0 - mcr p15, 0, r0, c7, c10, 4 @ drain write buffer mov pc, lr /* diff --git a/arch/arm/mm/cache-v4wb.S b/arch/arm/mm/cache-v4wb.S index f40c696..523c0cb 100644 --- a/arch/arm/mm/cache-v4wb.S +++ b/arch/arm/mm/cache-v4wb.S @@ -194,7 +194,6 @@ v4wb_dma_inv_range: add r0, r0, #CACHE_DLINESIZE cmp r0, r1 blo 1b - mcr p15, 0, r0, c7, c10, 4 @ drain write buffer mov pc, lr /* @@ -211,7 +210,6 @@ v4wb_dma_clean_range: add r0, r0, #CACHE_DLINESIZE cmp r0, r1 blo 1b - mcr p15, 0, r0, c7, c10, 4 @ drain write buffer mov pc, lr /* diff --git a/arch/arm/mm/cache-v6.S b/arch/arm/mm/cache-v6.S index 73b4a8b..7a842dd 100644 --- a/arch/arm/mm/cache-v6.S +++ b/arch/arm/mm/cache-v6.S @@ -239,8 +239,6 @@ v6_dma_inv_range: strlo r2, [r0] @ write for ownership #endif blo 1b - mov r0, #0 - mcr p15, 0, r0, c7, c10, 4 @ drain write buffer mov pc, lr /* @@ -262,8 +260,6 @@ v6_dma_clean_range: add r0, r0, #D_CACHE_LINE_SIZE cmp r0, r1 blo 1b - mov r0, #0 - mcr p15, 0, r0, c7, c10, 4 @ drain write buffer mov pc, lr /* @@ -290,8 +286,6 @@ ENTRY(v6_dma_flush_range) strlob r2, [r0] @ write for ownership #endif blo 1b - mov r0, #0 - mcr p15, 0, r0, c7, c10, 4 @ drain write buffer mov pc, lr /* diff --git a/arch/arm/mm/cache-v7.S b/arch/arm/mm/cache-v7.S index d32f02b..18dcef6 100644 --- a/arch/arm/mm/cache-v7.S +++ b/arch/arm/mm/cache-v7.S @@ -257,7 +257,6 @@ v7_dma_inv_range: add r0, r0, r2 cmp r0, r1 blo 1b - dsb mov pc, lr ENDPROC(v7_dma_clean_range) @@ -293,7 +291,6 @@ ENTRY(v7_dma_flush_range) add r0, r0, r2 cmp r0, r1 blo 1b - dsb mov pc, lr ENDPROC(v7_dma_flush_range) diff --git a/arch/arm/mm/dma-mapping.c b/arch/arm/mm/dma-mapping.c index 82a093c..ff85283 100644 --- a/arch/arm/mm/dma-mapping.c +++ b/arch/arm/mm/dma-mapping.c @@ -97,6 +97,7 @@ static struct page *__dma_alloc_buffer(struct device *dev, size_t size, gfp_t gf memset(ptr, 0, size); dmac_flush_range(ptr, ptr + size); outer_flush_range(__pa(ptr), __pa(ptr) + size); + __dma_sync(); return page;