[PATCH v2 09/10] iommu/arm-smmu-v3: Add types for each level of the CD table
Jason Gunthorpe
jgg at nvidia.com
Mon Jun 10 17:31:18 PDT 2024
As well as indexing helpers arm_smmu_cdtab_l1/2_idx().
Remove CTXDESC_L1_DESC_DWORDS and CTXDESC_CD_DWORDS replacing them all
with type specific calculations.
Signed-off-by: Jason Gunthorpe <jgg at nvidia.com>
---
drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c | 46 +++++++++++----------
drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.h | 27 +++++++++---
2 files changed, 46 insertions(+), 27 deletions(-)
diff --git a/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c b/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c
index dd65e27aebafd4..49e019131ef3b5 100644
--- a/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c
+++ b/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c
@@ -1167,17 +1167,18 @@ static void arm_smmu_sync_cd(struct arm_smmu_master *master,
arm_smmu_cmdq_batch_submit(smmu, &cmds);
}
-static void arm_smmu_write_cd_l1_desc(__le64 *dst, dma_addr_t l2ptr_dma)
+static void arm_smmu_write_cd_l1_desc(struct arm_smmu_cdtab_l1 *dst,
+ dma_addr_t l2ptr_dma)
{
u64 val = (l2ptr_dma & CTXDESC_L1_DESC_L2PTR_MASK) | CTXDESC_L1_DESC_V;
/* The HW has 64 bit atomicity with stores to the L2 CD table */
- WRITE_ONCE(*dst, cpu_to_le64(val));
+ WRITE_ONCE(dst->l2ptr, cpu_to_le64(val));
}
-static dma_addr_t arm_smmu_cd_l1_get_desc(const __le64 *src)
+static dma_addr_t arm_smmu_cd_l1_get_desc(const struct arm_smmu_cdtab_l1 *src)
{
- return le64_to_cpu(src) & CTXDESC_L1_DESC_L2PTR_MASK;
+ return le64_to_cpu(src->l2ptr) & CTXDESC_L1_DESC_L2PTR_MASK;
}
struct arm_smmu_cd *arm_smmu_get_cd_ptr(struct arm_smmu_master *master,
@@ -1190,13 +1191,12 @@ struct arm_smmu_cd *arm_smmu_get_cd_ptr(struct arm_smmu_master *master,
return NULL;
if (cd_table->s1fmt == STRTAB_STE_0_S1FMT_LINEAR)
- return (struct arm_smmu_cd *)(cd_table->cdtab +
- ssid * CTXDESC_CD_DWORDS);
+ return &((struct arm_smmu_cd *)cd_table->cdtab)[ssid];
- l1_desc = &cd_table->l1_desc[ssid / CTXDESC_L2_ENTRIES];
+ l1_desc = &cd_table->l1_desc[arm_smmu_cdtab_l1_idx(ssid)];
if (!l1_desc->l2ptr)
return NULL;
- return &l1_desc->l2ptr[ssid % CTXDESC_L2_ENTRIES];
+ return &l1_desc->l2ptr->cds[arm_smmu_cdtab_l2_idx(ssid)];
}
struct arm_smmu_cd *arm_smmu_alloc_cd_ptr(struct arm_smmu_master *master,
@@ -1214,7 +1214,7 @@ struct arm_smmu_cd *arm_smmu_alloc_cd_ptr(struct arm_smmu_master *master,
}
if (cd_table->s1fmt == STRTAB_STE_0_S1FMT_64K_L2) {
- unsigned int idx = ssid / CTXDESC_L2_ENTRIES;
+ unsigned int idx = arm_smmu_cdtab_l1_idx(ssid);
struct arm_smmu_l1_ctx_desc *l1_desc;
l1_desc = &cd_table->l1_desc[idx];
@@ -1228,8 +1228,10 @@ struct arm_smmu_cd *arm_smmu_alloc_cd_ptr(struct arm_smmu_master *master,
if (!l1_desc->l2ptr)
return NULL;
- arm_smmu_write_cd_l1_desc(&cd_table->cdtab[idx],
- l2ptr_dma);
+ arm_smmu_write_cd_l1_desc(
+ &((struct arm_smmu_cdtab_l1 *)
+ cd_table->cdtab)[idx],
+ l2ptr_dma);
/* An invalid L1CD can be cached */
arm_smmu_sync_cd(master, ssid, false);
}
@@ -1358,7 +1360,7 @@ static int arm_smmu_alloc_cd_tables(struct arm_smmu_master *master)
cd_table->s1fmt = STRTAB_STE_0_S1FMT_LINEAR;
cd_table->num_l1_ents = max_contexts;
- l1size = max_contexts * (CTXDESC_CD_DWORDS << 3);
+ l1size = max_contexts * sizeof(struct arm_smmu_cd);
} else {
cd_table->s1fmt = STRTAB_STE_0_S1FMT_64K_L2;
cd_table->num_l1_ents = DIV_ROUND_UP(max_contexts,
@@ -1370,7 +1372,8 @@ static int arm_smmu_alloc_cd_tables(struct arm_smmu_master *master)
if (!cd_table->l1_desc)
return -ENOMEM;
- l1size = cd_table->num_l1_ents * (CTXDESC_L1_DESC_DWORDS << 3);
+ l1size = cd_table->num_l1_ents *
+ sizeof(struct arm_smmu_cdtab_l1);
}
cd_table->cdtab = dma_alloc_coherent(smmu->dev, l1size,
@@ -1394,27 +1397,28 @@ static int arm_smmu_alloc_cd_tables(struct arm_smmu_master *master)
static void arm_smmu_free_cd_tables(struct arm_smmu_master *master)
{
int i;
- size_t size, l1size;
+ size_t l1size;
struct arm_smmu_device *smmu = master->smmu;
struct arm_smmu_ctx_desc_cfg *cd_table = &master->cd_table;
if (cd_table->l1_desc) {
- size = CTXDESC_L2_ENTRIES * (CTXDESC_CD_DWORDS << 3);
-
for (i = 0; i < cd_table->num_l1_ents; i++) {
if (!cd_table->l1_desc[i].l2ptr)
continue;
- dma_free_coherent(smmu->dev, size,
+ dma_free_coherent(smmu->dev,
+ sizeof(*cd_table->l1_desc[i].l2ptr),
cd_table->l1_desc[i].l2ptr,
- arm_smmu_cd_l1_get_desc(
- &cd_table->cdtab[i]));
+ arm_smmu_cd_l1_get_desc(&(
+ (struct arm_smmu_cdtab_l1 *)
+ cd_table->cdtab)[i]));
}
kfree(cd_table->l1_desc);
- l1size = cd_table->num_l1_ents * (CTXDESC_L1_DESC_DWORDS << 3);
+ l1size = cd_table->num_l1_ents *
+ sizeof(struct arm_smmu_cdtab_l1);
} else {
- l1size = cd_table->num_l1_ents * (CTXDESC_CD_DWORDS << 3);
+ l1size = cd_table->num_l1_ents * sizeof(struct arm_smmu_cd);
}
dma_free_coherent(smmu->dev, l1size, cd_table->cdtab,
diff --git a/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.h b/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.h
index 2f7d70d92b1f31..7ed8d53423cf56 100644
--- a/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.h
+++ b/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.h
@@ -293,16 +293,31 @@ static inline unsigned int arm_smmu_strtab_l2_idx(unsigned int sid)
*/
#define CTXDESC_L2_ENTRIES 1024
-#define CTXDESC_L1_DESC_DWORDS 1
#define CTXDESC_L1_DESC_V (1UL << 0)
#define CTXDESC_L1_DESC_L2PTR_MASK GENMASK_ULL(51, 12)
-#define CTXDESC_CD_DWORDS 8
-
struct arm_smmu_cd {
- __le64 data[CTXDESC_CD_DWORDS];
+ __le64 data[8];
};
+struct arm_smmu_cdtab_l2 {
+ struct arm_smmu_cd cds[CTXDESC_L2_ENTRIES];
+};
+
+struct arm_smmu_cdtab_l1 {
+ __le64 l2ptr;
+};
+
+static inline unsigned int arm_smmu_cdtab_l1_idx(unsigned int ssid)
+{
+ return ssid / CTXDESC_L2_ENTRIES;
+}
+
+static inline unsigned int arm_smmu_cdtab_l2_idx(unsigned int ssid)
+{
+ return ssid % CTXDESC_L2_ENTRIES;
+}
+
#define CTXDESC_CD_0_TCR_T0SZ GENMASK_ULL(5, 0)
#define CTXDESC_CD_0_TCR_TG0 GENMASK_ULL(7, 6)
#define CTXDESC_CD_0_TCR_IRGN0 GENMASK_ULL(9, 8)
@@ -330,7 +345,7 @@ struct arm_smmu_cd {
* When the SMMU only supports linear context descriptor tables, pick a
* reasonable size limit (64kB).
*/
-#define CTXDESC_LINEAR_CDMAX ilog2(SZ_64K / (CTXDESC_CD_DWORDS << 3))
+#define CTXDESC_LINEAR_CDMAX ilog2(SZ_64K / sizeof(struct arm_smmu_cd))
/* Command queue */
#define CMDQ_ENT_SZ_SHIFT 4
@@ -602,7 +617,7 @@ struct arm_smmu_ctx_desc {
};
struct arm_smmu_l1_ctx_desc {
- struct arm_smmu_cd *l2ptr;
+ struct arm_smmu_cdtab_l2 *l2ptr;
};
struct arm_smmu_ctx_desc_cfg {
--
2.45.2
More information about the linux-arm-kernel
mailing list