[PATCH v3 1/3] powerpc/pseries: Macros and wrapper functions for H_HTM call

[PATCH v3 1/3] powerpc/pseries: Macros and wrapper functions for H_HTM call

[Madhavan Srinivasan]

Define macros and wrapper functions to handle
H_HTM (Hardware Trace Macro) hypervisor call.
H_HTM is new HCALL added to export data from
Hardware Trace Macro (HTM) function.

Signed-off-by: Madhavan Srinivasan <maddy@linux.ibm.com>
---
Changelog v2:
- No changes
Changelog v1:
- No changes

 arch/powerpc/include/asm/hvcall.h         | 34 +++++++++++++++++++++++
 arch/powerpc/include/asm/plpar_wrappers.h | 21 ++++++++++++++
 2 files changed, 55 insertions(+)

diff --git a/arch/powerpc/include/asm/hvcall.h b/arch/powerpc/include/asm/hvcall.h
index 7a8495660c2f..7ad13685c127 100644
--- a/arch/powerpc/include/asm/hvcall.h
+++ b/arch/powerpc/include/asm/hvcall.h
@@ -348,6 +348,7 @@
 #define H_SCM_FLUSH		0x44C
 #define H_GET_ENERGY_SCALE_INFO	0x450
 #define H_PKS_SIGNED_UPDATE	0x454
+#define H_HTM			0x458
 #define H_WATCHDOG		0x45C
 #define H_GUEST_GET_CAPABILITIES 0x460
 #define H_GUEST_SET_CAPABILITIES 0x464
@@ -497,6 +498,39 @@
 #define H_GUEST_CAP_POWER10	(1UL<<(63-2))
 #define H_GUEST_CAP_BITMAP2	(1UL<<(63-63))
 
+/*
+ * Defines for H_HTM - Macros for hardware trace macro (HTM) function.
+ */
+#define H_HTM_FLAGS_HARDWARE_TARGET	(1ul << 63)
+#define H_HTM_FLAGS_LOGICAL_TARGET	(1ul << 62)
+#define H_HTM_FLAGS_PROCID_TARGET	(1ul << 61)
+#define H_HTM_FLAGS_NOWRAP		(1ul << 60)
+
+#define H_HTM_OP_SHIFT			(63-15)
+#define H_HTM_OP(x)			((unsigned long)(x)<<H_HTM_OP_SHIFT)
+#define H_HTM_OP_CAPABILITIES		0x01
+#define H_HTM_OP_STATUS			0x02
+#define H_HTM_OP_SETUP			0x03
+#define H_HTM_OP_CONFIGURE		0x04
+#define H_HTM_OP_START			0x05
+#define H_HTM_OP_STOP			0x06
+#define H_HTM_OP_DECONFIGURE		0x07
+#define H_HTM_OP_DUMP_DETAILS		0x08
+#define H_HTM_OP_DUMP_DATA		0x09
+#define H_HTM_OP_DUMP_SYSMEM_CONF	0x0a
+#define H_HTM_OP_DUMP_SYSPROC_CONF	0x0b
+
+#define H_HTM_TYPE_SHIFT		(63-31)
+#define H_HTM_TYPE(x)			((unsigned long)(x)<<H_HTM_TYPE_SHIFT)
+#define H_HTM_TYPE_NEST			0x01
+#define H_HTM_TYPE_CORE			0x02
+#define H_HTM_TYPE_LLAT			0x03
+#define H_HTM_TYPE_GLOBAL		0xff
+
+#define H_HTM_TARGET_NODE_INDEX(x)		((unsigned long)(x)<<(63-15))
+#define H_HTM_TARGET_NODAL_CHIP_INDEX(x)	((unsigned long)(x)<<(63-31))
+#define H_HTM_TARGET_CORE_INDEX_ON_CHIP(x)	((unsigned long)(x)<<(63-47))
+
 #ifndef __ASSEMBLY__
 #include <linux/types.h>
 
diff --git a/arch/powerpc/include/asm/plpar_wrappers.h b/arch/powerpc/include/asm/plpar_wrappers.h
index 71648c126970..a63a4d29cfdf 100644
--- a/arch/powerpc/include/asm/plpar_wrappers.h
+++ b/arch/powerpc/include/asm/plpar_wrappers.h
@@ -65,6 +65,27 @@ static inline long register_dtl(unsigned long cpu, unsigned long vpa)
 	return vpa_call(H_VPA_REG_DTL, cpu, vpa);
 }
 
+static inline long htm_call(unsigned long flags, unsigned long target,
+		unsigned long operation, unsigned long param1,
+		unsigned long param2, unsigned long param3)
+{
+	return plpar_hcall_norets(H_HTM, flags, target, operation,
+				  param1, param2, param3);
+}
+
+static inline long htm_get_dump_hardware(unsigned long nodeindex,
+		unsigned long nodalchipindex, unsigned long coreindexonchip,
+		unsigned long type, unsigned long addr, unsigned long size,
+		unsigned long offset)
+{
+	return htm_call(H_HTM_FLAGS_HARDWARE_TARGET,
+			H_HTM_TARGET_NODE_INDEX(nodeindex) |
+			H_HTM_TARGET_NODAL_CHIP_INDEX(nodalchipindex) |
+			H_HTM_TARGET_CORE_INDEX_ON_CHIP(coreindexonchip),
+			H_HTM_OP(H_HTM_OP_DUMP_DATA) | H_HTM_TYPE(type),
+			addr, size, offset);
+}
+
 extern void vpa_init(int cpu);
 
 static inline long plpar_pte_enter(unsigned long flags,
-- 
2.45.2

[PATCH v3 2/3] powerpc/pseries: Export hardware trace macro dump via debugfs

[Madhavan Srinivasan]

This patch adds debugfs interface to export Hardware Trace Macro (HTM)
function data in a LPAR. New hypervisor call "H_HTM" has been
defined to setup, configure, control and dump the HTM data.
This patch supports only dumping of HTM data in a LPAR.
New debugfs folder called "htmdump" has been added under
/sys/kernel/debug/arch path which contains files need to
pass required parameters for the H_HTM dump function. New Kconfig
option called "CONFIG_HTMDUMP" has been in platform/pseries for the same.

With this module loaded, list of files in debugfs path

/sys/kernel/debug/powerpc/htmdump
coreindexonchip  htmtype  nodalchipindex  nodeindex  trace

Signed-off-by: Madhavan Srinivasan <maddy@linux.ibm.com>
---
Changelog v2:
- Made driver as modules based on review comments
Changelog v1:
- Changed from tristate to bool with dependency flags
- Trimmed the include headers

 arch/powerpc/platforms/pseries/Kconfig   |   9 ++
 arch/powerpc/platforms/pseries/Makefile  |   1 +
 arch/powerpc/platforms/pseries/htmdump.c | 130 +++++++++++++++++++++++
 3 files changed, 140 insertions(+)
 create mode 100644 arch/powerpc/platforms/pseries/htmdump.c

diff --git a/arch/powerpc/platforms/pseries/Kconfig b/arch/powerpc/platforms/pseries/Kconfig
index afc0f6a61337..a66be66d690e 100644
--- a/arch/powerpc/platforms/pseries/Kconfig
+++ b/arch/powerpc/platforms/pseries/Kconfig
@@ -128,6 +128,15 @@ config CMM
 	  will be reused for other LPARs. The interface allows firmware to
 	  balance memory across many LPARs.
 
+config HTMDUMP
+	tristate "PHYP HTM data dumper"
+	depends on PPC_PSERIES && DEBUG_FS
+	default m
+	help
+	  Select this option, if you want to enable the kernel debugfs
+	  interface to dump the Hardware Trace Macro (HTM) function data
+	  in the LPAR.
+
 config HV_PERF_CTRS
 	bool "Hypervisor supplied PMU events (24x7 & GPCI)"
 	default y
diff --git a/arch/powerpc/platforms/pseries/Makefile b/arch/powerpc/platforms/pseries/Makefile
index 7bf506f6b8c8..3f3e3492e436 100644
--- a/arch/powerpc/platforms/pseries/Makefile
+++ b/arch/powerpc/platforms/pseries/Makefile
@@ -19,6 +19,7 @@ obj-$(CONFIG_HVC_CONSOLE)	+= hvconsole.o
 obj-$(CONFIG_HVCS)		+= hvcserver.o
 obj-$(CONFIG_HCALL_STATS)	+= hvCall_inst.o
 obj-$(CONFIG_CMM)		+= cmm.o
+obj-$(CONFIG_HTMDUMP)		+= htmdump.o
 obj-$(CONFIG_IO_EVENT_IRQ)	+= io_event_irq.o
 obj-$(CONFIG_LPARCFG)		+= lparcfg.o
 obj-$(CONFIG_IBMVIO)		+= vio.o
diff --git a/arch/powerpc/platforms/pseries/htmdump.c b/arch/powerpc/platforms/pseries/htmdump.c
new file mode 100644
index 000000000000..54c28525c4a7
--- /dev/null
+++ b/arch/powerpc/platforms/pseries/htmdump.c
@@ -0,0 +1,130 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Copyright (C) IBM Corporation, 2024
+ */
+
+#define pr_fmt(fmt) "htmdump: " fmt
+
+#include <linux/debugfs.h>
+
+#include <asm/machdep.h>
+#include <asm/plpar_wrappers.h>
+
+/* This enables us to keep track of the memory removed from each node. */
+struct htmdump_entry {
+	void *buf;
+	struct dentry *dir;
+	char name[16];
+};
+
+static u32 nodeindex;
+static u32 nodalchipindex;
+static u32 coreindexonchip;
+static u32 htmtype;
+static struct dentry *htmdump_debugfs_dir;
+static struct htmdump_entry *ent;
+
+#define BUFFER_SIZE PAGE_SIZE
+
+static ssize_t htmdump_read(struct file *filp, char __user *ubuf,
+			     size_t count, loff_t *ppos)
+{
+	struct htmdump_entry *ent = filp->private_data;
+	unsigned long page, read_size, available;
+	loff_t offset;
+	long rc;
+
+	page = ALIGN_DOWN(*ppos, BUFFER_SIZE);
+	offset = (*ppos) % BUFFER_SIZE;
+
+	rc = htm_get_dump_hardware(nodeindex, nodalchipindex, coreindexonchip,
+				   htmtype, virt_to_phys(ent->buf), BUFFER_SIZE, page);
+
+	switch (rc) {
+	case H_SUCCESS:
+	case H_PARTIAL:
+		break;
+	case H_NOT_AVAILABLE:
+		return 0;
+	case H_BUSY:
+	case H_LONG_BUSY_ORDER_1_MSEC:
+	case H_LONG_BUSY_ORDER_10_MSEC:
+	case H_LONG_BUSY_ORDER_100_MSEC:
+	case H_LONG_BUSY_ORDER_1_SEC:
+	case H_LONG_BUSY_ORDER_10_SEC:
+	case H_LONG_BUSY_ORDER_100_SEC:
+	case H_PARAMETER:
+	case H_P2:
+	case H_P3:
+	case H_P4:
+	case H_P5:
+	case H_P6:
+	case H_STATE:
+	case H_AUTHORITY:
+		return -EINVAL;
+	}
+
+	available = BUFFER_SIZE - offset;
+	read_size = min(count, available);
+	*ppos += read_size;
+	return simple_read_from_buffer(ubuf, count, &offset, ent->buf, available);
+}
+
+static const struct file_operations htmdump_fops = {
+	.llseek = default_llseek,
+	.read	= htmdump_read,
+	.open	= simple_open,
+};
+
+static int htmdump_init_debugfs(void)
+{
+	ent = kcalloc(1, sizeof(struct htmdump_entry), GFP_KERNEL);
+	if (!ent) {
+		pr_err("Failed to allocate ent\n");
+		return -EINVAL;
+	}
+
+	ent->buf = kmalloc(BUFFER_SIZE, GFP_KERNEL);
+	if (!ent->buf) {
+		pr_err("Failed to allocate htmdump buf\n");
+		return -ENOMEM;
+	}
+
+	pr_debug("%s: ent:%lx buf:%lx\n",
+			__func__, (long unsigned int)ent, (long unsigned int)ent->buf);
+
+	htmdump_debugfs_dir = debugfs_create_dir("htmdump",
+						  arch_debugfs_dir);
+
+	debugfs_create_u32("nodeindex", 0600,
+			htmdump_debugfs_dir, &nodeindex);
+	debugfs_create_u32("nodalchipindex", 0600,
+			htmdump_debugfs_dir, &nodalchipindex);
+	debugfs_create_u32("coreindexonchip", 0600,
+			htmdump_debugfs_dir, &coreindexonchip);
+	debugfs_create_u32("htmtype", 0600,
+			htmdump_debugfs_dir, &htmtype);
+	debugfs_create_file("trace", 0400, htmdump_debugfs_dir, ent, &htmdump_fops);
+
+	return 0;
+}
+
+static int __init htmdump_init(void)
+{
+	if (htmdump_init_debugfs())
+		return -EINVAL;
+
+	return 0;
+}
+
+static void __exit htmdump_exit(void)
+{
+	debugfs_remove_recursive(htmdump_debugfs_dir);
+	kfree(ent->buf);
+	kfree(ent);
+}
+
+module_init(htmdump_init);
+module_exit(htmdump_exit);
+MODULE_DESCRIPTION("PHYP Hardware Trace Macro (HTM) data dumper");
+MODULE_LICENSE("GPL");
-- 
2.45.2

Re: [PATCH v3 2/3] powerpc/pseries: Export hardware trace macro dump via debugfs

[Ritesh Harjani]

Madhavan Srinivasan <maddy@linux.ibm.com> writes:

> This patch adds debugfs interface to export Hardware Trace Macro (HTM)
> function data in a LPAR. New hypervisor call "H_HTM" has been
> defined to setup, configure, control and dump the HTM data.
> This patch supports only dumping of HTM data in a LPAR.
> New debugfs folder called "htmdump" has been added under
> /sys/kernel/debug/arch path which contains files need to
> pass required parameters for the H_HTM dump function. New Kconfig
> option called "CONFIG_HTMDUMP" has been in platform/pseries for the same.
>
> With this module loaded, list of files in debugfs path
>
> /sys/kernel/debug/powerpc/htmdump
> coreindexonchip  htmtype  nodalchipindex  nodeindex  trace
>
> Signed-off-by: Madhavan Srinivasan <maddy@linux.ibm.com>
> ---
> Changelog v2:
> - Made driver as modules based on review comments
> Changelog v1:
> - Changed from tristate to bool with dependency flags
> - Trimmed the include headers
>
>  arch/powerpc/platforms/pseries/Kconfig   |   9 ++
>  arch/powerpc/platforms/pseries/Makefile  |   1 +
>  arch/powerpc/platforms/pseries/htmdump.c | 130 +++++++++++++++++++++++
>  3 files changed, 140 insertions(+)
>  create mode 100644 arch/powerpc/platforms/pseries/htmdump.c
>
> diff --git a/arch/powerpc/platforms/pseries/Kconfig b/arch/powerpc/platforms/pseries/Kconfig
> index afc0f6a61337..a66be66d690e 100644
> --- a/arch/powerpc/platforms/pseries/Kconfig
> +++ b/arch/powerpc/platforms/pseries/Kconfig
> @@ -128,6 +128,15 @@ config CMM
>  	  will be reused for other LPARs. The interface allows firmware to
>  	  balance memory across many LPARs.
>  
> +config HTMDUMP
> +	tristate "PHYP HTM data dumper"
> +	depends on PPC_PSERIES && DEBUG_FS
> +	default m
> +	help
> +	  Select this option, if you want to enable the kernel debugfs
> +	  interface to dump the Hardware Trace Macro (HTM) function data
> +	  in the LPAR.
> +
>  config HV_PERF_CTRS
>  	bool "Hypervisor supplied PMU events (24x7 & GPCI)"
>  	default y
> diff --git a/arch/powerpc/platforms/pseries/Makefile b/arch/powerpc/platforms/pseries/Makefile
> index 7bf506f6b8c8..3f3e3492e436 100644
> --- a/arch/powerpc/platforms/pseries/Makefile
> +++ b/arch/powerpc/platforms/pseries/Makefile
> @@ -19,6 +19,7 @@ obj-$(CONFIG_HVC_CONSOLE)	+= hvconsole.o
>  obj-$(CONFIG_HVCS)		+= hvcserver.o
>  obj-$(CONFIG_HCALL_STATS)	+= hvCall_inst.o
>  obj-$(CONFIG_CMM)		+= cmm.o
> +obj-$(CONFIG_HTMDUMP)		+= htmdump.o
>  obj-$(CONFIG_IO_EVENT_IRQ)	+= io_event_irq.o
>  obj-$(CONFIG_LPARCFG)		+= lparcfg.o
>  obj-$(CONFIG_IBMVIO)		+= vio.o
> diff --git a/arch/powerpc/platforms/pseries/htmdump.c b/arch/powerpc/platforms/pseries/htmdump.c
> new file mode 100644
> index 000000000000..54c28525c4a7
> --- /dev/null
> +++ b/arch/powerpc/platforms/pseries/htmdump.c
> @@ -0,0 +1,130 @@
> +// SPDX-License-Identifier: GPL-2.0-or-later
> +/*
> + * Copyright (C) IBM Corporation, 2024
> + */
> +
> +#define pr_fmt(fmt) "htmdump: " fmt
> +
> +#include <linux/debugfs.h>
> +
> +#include <asm/machdep.h>
> +#include <asm/plpar_wrappers.h>
> +
> +/* This enables us to keep track of the memory removed from each node. */
> +struct htmdump_entry {
> +	void *buf;
> +	struct dentry *dir;
> +	char name[16];
> +};
>

How does dir and name gets used?
It isn't that obvious, so maybe a comment will be gr8.

> +static u32 nodeindex;
> +static u32 nodalchipindex;
> +static u32 coreindexonchip;
> +static u32 htmtype;
> +static struct dentry *htmdump_debugfs_dir;
> +static struct htmdump_entry *ent;
> +
> +#define BUFFER_SIZE PAGE_SIZE
> +
> +static ssize_t htmdump_read(struct file *filp, char __user *ubuf,
> +			     size_t count, loff_t *ppos)
> +{
> +	struct htmdump_entry *ent = filp->private_data;
> +	unsigned long page, read_size, available;
> +	loff_t offset;
> +	long rc;
> +
> +	page = ALIGN_DOWN(*ppos, BUFFER_SIZE);
> +	offset = (*ppos) % BUFFER_SIZE;
> +
> +	rc = htm_get_dump_hardware(nodeindex, nodalchipindex, coreindexonchip,
> +				   htmtype, virt_to_phys(ent->buf), BUFFER_SIZE, page);
> +
> +	switch (rc) {
> +	case H_SUCCESS:
> +	case H_PARTIAL:
> +		break;
> +	case H_NOT_AVAILABLE:
> +		return 0;

Minor nits for error returns here...

Is returning 0 correct here? Maybe it is (since 0 means no data read),
but wanted to confirm if we should return -ENODATA, or -ENODEV 
(not sure what does H_NOT_AVAILABLE here means)

#define	ENODATA		61	/* No data available */

> +	case H_BUSY:
> +	case H_LONG_BUSY_ORDER_1_MSEC:
> +	case H_LONG_BUSY_ORDER_10_MSEC:
> +	case H_LONG_BUSY_ORDER_100_MSEC:
> +	case H_LONG_BUSY_ORDER_1_SEC:
> +	case H_LONG_BUSY_ORDER_10_SEC:
> +	case H_LONG_BUSY_ORDER_100_SEC:

Similarly for above maybe -EBUSY perhaps, instead of -EINVAL

#define	EBUSY		16	/* Device or resource busy */

> +	case H_PARAMETER:
> +	case H_P2:
> +	case H_P3:
> +	case H_P4:
> +	case H_P5:
> +	case H_P6:
> +	case H_STATE:
> +	case H_AUTHORITY:
> +		return -EINVAL;
> +	}
> +
> +	available = BUFFER_SIZE - offset;
> +	read_size = min(count, available);
> +	*ppos += read_size;
> +	return simple_read_from_buffer(ubuf, count, &offset, ent->buf, available);
> +}
> +
> +static const struct file_operations htmdump_fops = {
> +	.llseek = default_llseek,
> +	.read	= htmdump_read,
> +	.open	= simple_open,
> +};
> +
> +static int htmdump_init_debugfs(void)
> +{
> +	ent = kcalloc(1, sizeof(struct htmdump_entry), GFP_KERNEL);
> +	if (!ent) {
> +		pr_err("Failed to allocate ent\n");
> +		return -EINVAL;

return value can be -ENOMEM;

> +	}
> +
> +	ent->buf = kmalloc(BUFFER_SIZE, GFP_KERNEL);
> +	if (!ent->buf) {
> +		pr_err("Failed to allocate htmdump buf\n");
> +		return -ENOMEM;

kfree(ent)?


> +	}
> +
> +	pr_debug("%s: ent:%lx buf:%lx\n",
> +			__func__, (long unsigned int)ent, (long unsigned int)ent->buf);
> +

maybe %p perhaps? 


<Documentation/core-api/printk-formats.rst>
Plain Pointers
--------------

::

	%p	abcdef12 or 00000000abcdef12

Pointers printed without a specifier extension (i.e unadorned %p) are
hashed to prevent leaking information about the kernel memory layout. This
has the added benefit of providing a unique identifier.
<...>
...the aim of printing the address is to provide
more information for debugging, use %p and boot the kernel with the
``no_hash_pointers`` parameter during debugging, which will print all %p
addresses unmodified. If you *really* always want the unmodified address, see
%px below.


> +	htmdump_debugfs_dir = debugfs_create_dir("htmdump",
> +						  arch_debugfs_dir);
> +
> +	debugfs_create_u32("nodeindex", 0600,
> +			htmdump_debugfs_dir, &nodeindex);
> +	debugfs_create_u32("nodalchipindex", 0600,
> +			htmdump_debugfs_dir, &nodalchipindex);
> +	debugfs_create_u32("coreindexonchip", 0600,
> +			htmdump_debugfs_dir, &coreindexonchip);
> +	debugfs_create_u32("htmtype", 0600,
> +			htmdump_debugfs_dir, &htmtype);
> +	debugfs_create_file("trace", 0400, htmdump_debugfs_dir, ent, &htmdump_fops);
> +
> +	return 0;
> +}
> +
> +static int __init htmdump_init(void)
> +{
> +	if (htmdump_init_debugfs())
> +		return -EINVAL;
> +
> +	return 0;
> +}
> +
> +static void __exit htmdump_exit(void)
> +{
> +	debugfs_remove_recursive(htmdump_debugfs_dir);
> +	kfree(ent->buf);
> +	kfree(ent);
> +}
> +
> +module_init(htmdump_init);
> +module_exit(htmdump_exit);
> +MODULE_DESCRIPTION("PHYP Hardware Trace Macro (HTM) data dumper");
> +MODULE_LICENSE("GPL");
> -- 
> 2.45.2

[PATCH v3 3/3] powerpc: Document details on H_HTM hcall

[Madhavan Srinivasan]

Add documentation to 'papr_hcalls.rst' describing the
input, output and return values of the H_HTM hcall as
per the internal specification.

Signed-off-by: Madhavan Srinivasan <maddy@linux.ibm.com>
---
Changelog v2:
- No changes
Changelog v1:
- Updated commit message to include htmdump folder files 

 Documentation/arch/powerpc/papr_hcalls.rst | 11 +++++++++++
 1 file changed, 11 insertions(+)

diff --git a/Documentation/arch/powerpc/papr_hcalls.rst b/Documentation/arch/powerpc/papr_hcalls.rst
index 80d2c0aadab5..805e1cb9bab9 100644
--- a/Documentation/arch/powerpc/papr_hcalls.rst
+++ b/Documentation/arch/powerpc/papr_hcalls.rst
@@ -289,6 +289,17 @@ to be issued multiple times in order to be completely serviced. The
 subsequent hcalls to the hypervisor until the hcall is completely serviced
 at which point H_SUCCESS or other error is returned by the hypervisor.
 
+**H_HTM**
+
+| Input: flags, target, operation (op), op-param1, op-param2, op-param3
+| Out: *dumphtmbufferdata*
+| Return Value: *H_Success,H_Busy,H_LongBusyOrder,H_Partial,H_Parameter,
+		 H_P2,H_P3,H_P4,H_P5,H_P6,H_State,H_Not_Available,H_Authority*
+
+H_HTM supports setup, configuration, control and dumping of Hardware Trace
+Macro (HTM) function and its data. HTM buffer stores tracing data for functions
+like core instruction, core LLAT and nest.
+
 References
 ==========
 .. [1] "Power Architecture Platform Reference"
-- 
2.45.2