[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <1456236764-1569-5-git-send-email-Suravee.Suthikulpanit@amd.com>
Date: Tue, 23 Feb 2016 08:12:38 -0600
From: Suravee Suthikulpanit <Suravee.Suthikulpanit@....com>
To: <joro@...tes.org>, <bp@...en8.de>, <peterz@...radead.org>,
<mingo@...hat.com>, <acme@...nel.org>
CC: <andihartmann@...enet.de>, <vw@...mu.org>, <labbott@...hat.com>,
<linux-kernel@...r.kernel.org>, <iommu@...ts.linux-foundation.org>,
"Suravee Suthikulpanit" <Suravee.Suthikulpanit@....com>
Subject: [PATCH V5 04/10] perf/amd/iommu: Modify IOMMU API to allow specifying IOMMU index
The current amd_iommu_pc_set_reg_val() cann not support multi-IOMMU.
So, this patch rename and modifies them to allow callers to specify
IOMMU index.
The function amd_iommu_pc_get_set_reg_val() is too confusing,
and does not support multi-IOMMU. So, this patch breaks it down
to amd_iommu_pc_[get|set]_counter(), and modifies them to allow
callers to specify IOMMU index.
Signed-off-by: Suravee Suthikulpanit <Suravee.Suthikulpanit@....com>
---
arch/x86/events/amd/iommu.c | 45 +++++++++++------------
arch/x86/include/asm/perf/amd/iommu.h | 8 +++--
drivers/iommu/amd_iommu_init.c | 67 ++++++++++++++++++++++++++++++-----
3 files changed, 84 insertions(+), 36 deletions(-)
diff --git a/arch/x86/events/amd/iommu.c b/arch/x86/events/amd/iommu.c
index 67ff3f1..8a29754 100644
--- a/arch/x86/events/amd/iommu.c
+++ b/arch/x86/events/amd/iommu.c
@@ -243,46 +243,44 @@ static void perf_iommu_enable_event(struct perf_event *ev)
{
u8 csource = _GET_CSOURCE(ev);
u16 devid = _GET_DEVID(ev);
+ u8 bank = _GET_BANK(ev);
+ u8 cntr = _GET_CNTR(ev);
u64 reg = 0ULL;
reg = csource;
- amd_iommu_pc_get_set_reg_val(devid,
- _GET_BANK(ev), _GET_CNTR(ev) ,
- IOMMU_PC_COUNTER_SRC_REG, ®, true);
+ amd_iommu_pc_set_reg(0, devid, bank, cntr,
+ IOMMU_PC_COUNTER_SRC_REG, ®);
reg = 0ULL | devid | (_GET_DEVID_MASK(ev) << 32);
if (reg)
reg |= (1UL << 31);
- amd_iommu_pc_get_set_reg_val(devid,
- _GET_BANK(ev), _GET_CNTR(ev) ,
- IOMMU_PC_DEVID_MATCH_REG, ®, true);
+ amd_iommu_pc_set_reg(0, devid, bank, cntr,
+ IOMMU_PC_DEVID_MATCH_REG, ®);
reg = 0ULL | _GET_PASID(ev) | (_GET_PASID_MASK(ev) << 32);
if (reg)
reg |= (1UL << 31);
- amd_iommu_pc_get_set_reg_val(devid,
- _GET_BANK(ev), _GET_CNTR(ev) ,
- IOMMU_PC_PASID_MATCH_REG, ®, true);
+ amd_iommu_pc_set_reg(0, devid, bank, cntr,
+ IOMMU_PC_PASID_MATCH_REG, ®);
reg = 0ULL | _GET_DOMID(ev) | (_GET_DOMID_MASK(ev) << 32);
if (reg)
reg |= (1UL << 31);
- amd_iommu_pc_get_set_reg_val(devid,
- _GET_BANK(ev), _GET_CNTR(ev) ,
- IOMMU_PC_DOMID_MATCH_REG, ®, true);
+ amd_iommu_pc_set_reg(0, devid, bank, cntr,
+ IOMMU_PC_DOMID_MATCH_REG, ®);
}
static void perf_iommu_disable_event(struct perf_event *event)
{
u64 reg = 0ULL;
- amd_iommu_pc_get_set_reg_val(_GET_DEVID(event),
- _GET_BANK(event), _GET_CNTR(event),
- IOMMU_PC_COUNTER_SRC_REG, ®, true);
+ amd_iommu_pc_set_reg(0, _GET_DEVID(event), _GET_BANK(event),
+ _GET_CNTR(event), IOMMU_PC_COUNTER_SRC_REG, ®);
}
static void perf_iommu_start(struct perf_event *event, int flags)
{
+ u64 val;
struct hw_perf_event *hwc = &event->hw;
pr_debug("perf: amd_iommu:perf_iommu_start\n");
@@ -292,13 +290,13 @@ static void perf_iommu_start(struct perf_event *event, int flags)
WARN_ON_ONCE(!(hwc->state & PERF_HES_UPTODATE));
hwc->state = 0;
- if (flags & PERF_EF_RELOAD) {
- u64 prev_raw_count = local64_read(&hwc->prev_count);
- amd_iommu_pc_get_set_reg_val(_GET_DEVID(event),
- _GET_BANK(event), _GET_CNTR(event),
- IOMMU_PC_COUNTER_REG, &prev_raw_count, true);
- }
+ if (!(flags & PERF_EF_RELOAD))
+ goto enable;
+
+ val = local64_read(&hwc->prev_count);
+ amd_iommu_pc_set_counter(0, _GET_BANK(event), _GET_CNTR(event), &val);
+enable:
perf_iommu_enable_event(event);
perf_event_update_userpage(event);
@@ -311,9 +309,8 @@ static void perf_iommu_read(struct perf_event *event)
struct hw_perf_event *hwc = &event->hw;
pr_debug("perf: amd_iommu:perf_iommu_read\n");
- amd_iommu_pc_get_set_reg_val(_GET_DEVID(event),
- _GET_BANK(event), _GET_CNTR(event),
- IOMMU_PC_COUNTER_REG, &count, false);
+ if (amd_iommu_pc_get_counter(0, _GET_BANK(event), _GET_CNTR(event), &cnt))
+ return;
/* IOMMU pc counter register is only 48 bits */
cnt &= GENMASK_ULL(48, 0);
diff --git a/arch/x86/include/asm/perf/amd/iommu.h b/arch/x86/include/asm/perf/amd/iommu.h
index 466be63..60ddfb6 100644
--- a/arch/x86/include/asm/perf/amd/iommu.h
+++ b/arch/x86/include/asm/perf/amd/iommu.h
@@ -31,9 +31,11 @@ u8 amd_iommu_pc_get_max_banks(int idx);
u8 amd_iommu_pc_get_max_counters(int idx);
-int amd_iommu_pc_set_reg_val(u16 devid, u8 bank, u8 cntr, u8 fxn, u64 *value);
+int amd_iommu_pc_set_reg(int idx, u16 devid, u8 bank, u8 cntr,
+ u8 fxn, u64 *value);
-int amd_iommu_pc_get_set_reg_val(u16 devid, u8 bank, u8 cntr, u8 fxn,
- u64 *value, bool is_write);
+int amd_iommu_pc_set_counter(int idx, u8 bank, u8 cntr, u64 *value);
+
+int amd_iommu_pc_get_counter(int idx, u8 bank, u8 cntr, u64 *value);
#endif /*_PERF_EVENT_AMD_IOMMU_H_*/
diff --git a/drivers/iommu/amd_iommu_init.c b/drivers/iommu/amd_iommu_init.c
index dde960f..b68a2d6 100644
--- a/drivers/iommu/amd_iommu_init.c
+++ b/drivers/iommu/amd_iommu_init.c
@@ -1133,6 +1133,8 @@ static int __init init_iommu_all(struct acpi_table_header *table)
return 0;
}
+static int iommu_pc_get_set_reg(struct amd_iommu *iommu, u8 bank, u8 cntr,
+ u8 fxn, u64 *value, bool is_write);
static void init_iommu_perf_ctr(struct amd_iommu *iommu)
{
@@ -1144,8 +1146,8 @@ static void init_iommu_perf_ctr(struct amd_iommu *iommu)
amd_iommu_pc_present = true;
/* Check if the performance counters can be written to */
- if ((0 != amd_iommu_pc_get_set_reg_val(0, 0, 0, 0, &val, true)) ||
- (0 != amd_iommu_pc_get_set_reg_val(0, 0, 0, 0, &val2, false)) ||
+ if ((iommu_pc_get_set_reg(iommu, 0, 0, 0, &val, true)) ||
+ (iommu_pc_get_set_reg(iommu, 0, 0, 0, &val2, false)) ||
(val != val2)) {
pr_err("AMD-Vi: Unable to write to IOMMU perf counter.\n");
amd_iommu_pc_present = false;
@@ -2292,10 +2294,9 @@ u8 amd_iommu_pc_get_max_counters(int idx)
}
EXPORT_SYMBOL(amd_iommu_pc_get_max_counters);
-int amd_iommu_pc_get_set_reg_val(u16 devid, u8 bank, u8 cntr, u8 fxn,
- u64 *value, bool is_write)
+static int iommu_pc_get_set_reg(struct amd_iommu *iommu, u8 bank, u8 cntr,
+ u8 fxn, u64 *value, bool is_write)
{
- struct amd_iommu *iommu;
u32 offset;
u32 max_offset_lim;
@@ -2303,9 +2304,6 @@ int amd_iommu_pc_get_set_reg_val(u16 devid, u8 bank, u8 cntr, u8 fxn,
if (!amd_iommu_pc_present)
return -ENODEV;
- /* Locate the iommu associated with the device ID */
- iommu = amd_iommu_rlookup_table[devid];
-
/* Check for valid iommu and pc register indexing */
if (WARN_ON((iommu == NULL) || (fxn > 0x28) || (fxn & 7)))
return -ENODEV;
@@ -2330,4 +2328,55 @@ int amd_iommu_pc_get_set_reg_val(u16 devid, u8 bank, u8 cntr, u8 fxn,
return 0;
}
-EXPORT_SYMBOL(amd_iommu_pc_get_set_reg_val);
+
+int amd_iommu_pc_set_reg(int idx, u16 devid, u8 bank, u8 cntr,
+ u8 fxn, u64 *value)
+{
+ struct amd_iommu *iommu = get_amd_iommu(idx);
+
+ if (!iommu)
+ return -EINVAL;
+
+ return iommu_pc_get_set_reg(iommu, bank, cntr, fxn, value, true);
+}
+EXPORT_SYMBOL(amd_iommu_pc_set_reg);
+
+int amd_iommu_pc_set_counter(int idx, u8 bank, u8 cntr, u64 *value)
+{
+ struct amd_iommu *iommu = get_amd_iommu(idx);
+
+ if (!iommu)
+ return -EINVAL;
+
+ return iommu_pc_get_set_reg(iommu, bank, cntr,
+ IOMMU_PC_COUNTER_REG,
+ value, true);
+}
+EXPORT_SYMBOL(amd_iommu_pc_set_counter);
+
+int amd_iommu_pc_get_counter(int idx, u8 bank, u8 cntr, u64 *value)
+{
+ struct amd_iommu *iommu = get_amd_iommu(idx);
+ int ret;
+ u64 tmp;
+
+ if (!value || !iommu)
+ return -EINVAL;
+ /*
+ * Here, we read the specified counters on all IOMMUs,
+ * which should have been programmed the same way and
+ * aggregate the counter values.
+ */
+
+ ret = iommu_pc_get_set_reg(iommu, bank, cntr,
+ IOMMU_PC_COUNTER_REG,
+ &tmp, false);
+ if (ret)
+ return ret;
+
+ /* IOMMU pc counter register is only 48 bits */
+ *value = tmp & GENMASK_ULL(48, 0);
+
+ return 0;
+}
+EXPORT_SYMBOL(amd_iommu_pc_get_counter);
--
1.9.1
Powered by blists - more mailing lists