Message ID | BANLkTi=vd2TE6JjueGmY=ydRQRygo8-Q1Q@mail.gmail.com (mailing list archive) |
---|---|
State | New, archived |
Headers | show |
* Ohad Ben-Cohen <ohad@wizery.com> [110630 23:47]: > Hi Tony, > > Please pull from: > > git://github.com/ohadbc/omap-iommu.git for-tony > > To receive trivial iommu/iovmm fixes (iommu clk name fix, pte fix and > a fix for iovmm's erroneous > usage of sg_dma_len as reported by Russell). > > All three patches were submitted to linux-omap and linux-arm-kernel for review, > and are also attached below for convenience. These look like real fixes, but can you please update the commits with something like "This fixes an issue with foo when doing bar". Otherwise it's easy to claim that these are "fixes for features that never worked" and it might be hard to justify them this late into the -rc cycle. Tony
On Fri, Jul 1, 2011 at 10:00 AM, Tony Lindgren <tony@atomide.com> wrote: > These look like real fixes, but can you please update the commits > with something like "This fixes an issue with foo when doing bar". Done, please tell me if it's OK now. > Otherwise it's easy to claim that these are "fixes for features that > never worked" and it might be hard to justify them this late into > the -rc cycle. It seems we did have these issues for quite some time, and these are not explicit 3.0 regressions. So they might vaguely fall into the "fixes for features that never worked" definition too in a sense. If you prefer to take them into 3.1, that's OK too. Thanks, Ohad.
* Ohad Ben-Cohen <ohad@wizery.com> [110701 00:46]: > On Fri, Jul 1, 2011 at 10:00 AM, Tony Lindgren <tony@atomide.com> wrote: > > These look like real fixes, but can you please update the commits > > with something like "This fixes an issue with foo when doing bar". > > Done, please tell me if it's OK now. OK thanks! > > Otherwise it's easy to claim that these are "fixes for features that > > never worked" and it might be hard to justify them this late into > > the -rc cycle. > > It seems we did have these issues for quite some time, and these are not > explicit 3.0 regressions. So they might vaguely fall into the "fixes > for features that never worked" definition too in a sense. > > If you prefer to take them into 3.1, that's OK too. Let's do that then. Regards, Tony
diff --git a/arch/arm/mach-omap2/iommu2.c b/arch/arm/mach-omap2/iommu2.c index adb083e..f286012 100644 --- a/arch/arm/mach-omap2/iommu2.c +++ b/arch/arm/mach-omap2/iommu2.c @@ -225,8 +225,8 @@ static u32 omap2_get_pte_attr(struct iotlb_entry *e) attr = e->mixed << 5; attr |= e->endian; attr |= e->elsz >> 3; - attr <<= ((e->pgsz & MMU_CAM_PGSZ_4K) ? 0 : 6); - + attr <<= (((e->pgsz == MMU_CAM_PGSZ_4K) || + (e->pgsz == MMU_CAM_PGSZ_64K)) ? 0 : 6); return attr; } diff --git a/arch/arm/mach-omap2/omap-iommu.c b/arch/arm/mach-omap2/omap-iommu.c index 3fc5dc7..e61fead 100644 --- a/arch/arm/mach-omap2/omap-iommu.c +++ b/arch/arm/mach-omap2/omap-iommu.c @@ -67,7 +67,7 @@ static struct iommu_device omap4_devices[] = { .pdata = { .name = "ducati", .nr_tlb_entries = 32, - .clk_name = "ducati_ick", + .clk_name = "ipu_fck", .da_start = 0x0, .da_end = 0xFFFFF000, }, diff --git a/arch/arm/plat-omap/iovmm.c b/arch/arm/plat-omap/iovmm.c index 83a37c5..c60737c 100644 --- a/arch/arm/plat-omap/iovmm.c +++ b/arch/arm/plat-omap/iovmm.c @@ -72,7 +72,7 @@ static size_t sgtable_len(const struct sg_table *sgt) for_each_sg(sgt->sgl, sg, sgt->nents, i) { size_t bytes; - bytes = sg_dma_len(sg); + bytes = sg->length; if (!iopgsz_ok(bytes)) { pr_err("%s: sg[%d] not iommu pagesize(%x)\n", @@ -198,7 +198,7 @@ static void *vmap_sg(const struct sg_table *sgt) int err; pa = sg_phys(sg); - bytes = sg_dma_len(sg); + bytes = sg->length; BUG_ON(bytes != PAGE_SIZE); @@ -476,7 +476,7 @@ static int map_iovm_area(struct iommu *obj, struct iovm_struct *new, struct iotlb_entry e; pa = sg_phys(sg); - bytes = sg_dma_len(sg); + bytes = sg->length; flags &= ~IOVMF_PGSZ_MASK;