Skip to contentSkip to conversation

Adding soc related info into /proc/cpuinfo

PATCH RESEND Superseded Eduardo Valentin 2 participants 7 messages May 5, 2010 – May 10, 2010

Subsystems (5)
Documentation Arm (32-bit) Arm sub-architectures OMAP2+ support +1 more subsystems
lore
Share & tools
lore

Eduardo Valentin

author
#0 lore raw

v3 cover Adding soc related info into /proc/cpuinfo

Hello all,

I'm resending this series with minor change in the subject prefix of patch 4/4. Removed the PM: prefix. Everything else is same. So, I kept the v3 prefix as well.

v3:
And now v3 of this series. Basically a minor change wrt string manipulation. No need to use strlen in so many places. Previous log for reference:

v2:

Here is the version 2 of this series. Now die id is protected using same x86 protection mechanism to hide x86 product number. Besides, a compilation Kconfig option has been added for DIE ID as well.

Here is previous PATCH 0/4 message:

v1:

This series is to continue what has been discussed several weeks ago wrt IDCODE patch. Original patch was made by Peter and discussion is here:
http://www.mail-archive.com/[email protected]/msg17553.html

So, the conclusion was that IDCODE info is useful even in production systems, and for that debugfs is not a good choice to export it. One suggestion was to add it under /proc/cpuinfo. However this entry nowadays exports only ARM related information.

So this series does the trick by extending the ARM /proc/cpuinfo to include soc info data. There are a few ways to add a hook for soc specific data. But I've decided to implement it via the simplest way I found. Basically it is same thing which is done for system_rev, system_serial_low and system_serial_high. Then, now we have system_soc_info, which is printed only if there is something useful there.

As usual, comments are welcome.

BR,

*** SUBJECT HERE ***

*** BLURB HERE ***

Eduardo Valentin (4):
ARM: Introduce SoC Info into /proc/cpuinfo
mach-omap2: Add SoC info data for OMAP2,3,4 into /proc/cpuinfo
mach-omap1: Add SoC info data for OMAP1 into /proc/cpuinfo
OMAP3: export chip IDCODE, Production ID and Die ID

 Documentation/kernel-parameters.txt |    2 +
 arch/arm/include/asm/system.h       |    2 +
 arch/arm/kernel/setup.c             |    5 +++
 arch/arm/mach-omap1/id.c            |   15 ++++++----
 arch/arm/mach-omap2/Kconfig         |   10 +++++++
 arch/arm/mach-omap2/id.c            |   51 ++++++++++++++++++++++++++++++----
 6 files changed, 73 insertions(+), 12 deletions(-)
author
#1 lore raw

PATCH RESEND v3 1/4 ARM: Introduce SoC Info into /proc/cpuinfo

From: Eduardo Valentin <[email protected]>

This patch extends the ARM /proc/cpuinfo to include soc info data. It is implemented via the same way which is done for system_rev, system_serial_low and system_serial_high. Then, now we have system_soc_info, which is printed only if there is something useful there.

Signed-off-by Eduardo Valentin <[email protected]>
---
 arch/arm/include/asm/system.h |    2 ++
 arch/arm/kernel/setup.c       |    5 +++++
 2 files changed, 7 insertions(+), 0 deletions(-)
arch/arm/include/asm/system.h +2 -0
index 4ace45e..53a9645 100644
@@ -71,6 +71,8 @@ struct task_struct;
 extern unsigned int system_rev;
 extern unsigned int system_serial_low;
 extern unsigned int system_serial_high;
+#define SYSTEM_SOC_INFO_SIZE			128
+extern char system_soc_info[SYSTEM_SOC_INFO_SIZE];
 extern unsigned int mem_fclk_21285;
 
 struct pt_regs;
arch/arm/kernel/setup.c +5 -0
index c91c77b..025d795 100644
@@ -85,6 +85,9 @@ EXPORT_SYMBOL(system_serial_low);
 unsigned int system_serial_high;
 EXPORT_SYMBOL(system_serial_high);
 
+char system_soc_info[SYSTEM_SOC_INFO_SIZE];
+EXPORT_SYMBOL(system_soc_info);
+
 unsigned int elf_hwcap;
 EXPORT_SYMBOL(elf_hwcap);
 

@@ -847,6 +850,8 @@ static int c_show(struct seq_file *m, void *v)
 	seq_printf(m, "Revision\t: %04x\n", system_rev);
 	seq_printf(m, "Serial\t\t: %08x%08x\n",
 		   system_serial_high, system_serial_low);
+	if (strlen(system_soc_info))
+		seq_printf(m, "SoC Info\t: %s\n",  system_soc_info);
 
 	return 0;
 }
Signature · 1 line
1.7.0.4.361.g8b5fe.dirty
author
#2 lore raw

PATCH RESEND v3 2/4 mach-omap2: Add SoC info data for OMAP2, 3, 4 into /proc/cpuinfo

In reply to Eduardo Valentin

From: Eduardo Valentin <[email protected]>

Report OMAP2,3,4 data into system_soc_info. Now we get omap information under /proc/cpuinfo.

Signed-off-by Eduardo Valentin <[email protected]>
---
 arch/arm/mach-omap2/id.c |   18 ++++++++++++------
 1 files changed, 12 insertions(+), 6 deletions(-)
arch/arm/mach-omap2/id.c +12 -6
index 37b8a1a..4702ffe 100644
@@ -104,7 +104,7 @@ static u16 tap_prod_id;
 
 void __init omap24xx_check_revision(void)
 {
-	int i, j;
+	int i, j, sz;
 	u32 idcode, prod_id;
 	u16 hawkeye;
 	u8  dev_type, rev;

@@ -152,10 +152,12 @@ void __init omap24xx_check_revision(void)
 		j = i;
 	}
 
-	pr_info("OMAP%04x", omap_rev() >> 16);
+	sz = snprintf(system_soc_info, SYSTEM_SOC_INFO_SIZE, "OMAP%04x",
+							omap_rev() >> 16);
 	if ((omap_rev() >> 8) & 0x0f)
-		pr_info("ES%x", (omap_rev() >> 12) & 0xf);
-	pr_info("\n");
+		snprintf(system_soc_info + sz, SYSTEM_SOC_INFO_SIZE - sz,
+				"ES%x",	(omap_rev() >> 12) & 0xf);
+	pr_info("%s\n", system_soc_info);
 }
 
 #define OMAP3_CHECK_FEATURE(status,feat)				\

@@ -286,7 +288,9 @@ void __init omap4_check_revision(void)
 	if ((hawkeye == 0xb852) && (rev == 0x0)) {
 		omap_revision = OMAP4430_REV_ES1_0;
 		omap_chip.oc |= CHIP_IS_OMAP4430ES1;
-		pr_info("OMAP%04x %s\n", omap_rev() >> 16, rev_name);
+		snprintf(system_soc_info, SYSTEM_SOC_INFO_SIZE, "OMAP%04x %s\n",
+				omap_rev() >> 16, rev_name);
+		pr_info("%s\n", system_soc_info);
 		return;
 	}
 

@@ -356,7 +360,9 @@ void __init omap3_cpuinfo(void)
 	}
 
 	/* Print verbose information */
-	pr_info("%s ES%s (", cpu_name, cpu_rev);
+	snprintf(system_soc_info, SYSTEM_SOC_INFO_SIZE, "%s ES%s", cpu_name,
+			cpu_rev);
+	pr_info("%s (", system_soc_info);
 
 	OMAP3_SHOW_FEATURE(l2cache);
 	OMAP3_SHOW_FEATURE(iva);
Signature · 1 line
1.7.0.4.361.g8b5fe.dirty
author
#3 lore raw

PATCH RESEND v3 3/4 mach-omap1: Add SoC info data for OMAP1 into /proc/cpuinfo

In reply to Eduardo Valentin

From: Eduardo Valentin <[email protected]>

Report OMAP data into system_soc_info. Now we get omap information under /proc/cpuinfo.

Signed-off-by Eduardo Valentin <[email protected]>
---
 arch/arm/mach-omap1/id.c |   15 +++++++++------
 1 files changed, 9 insertions(+), 6 deletions(-)
arch/arm/mach-omap1/id.c +9 -6
index a0e3560..9a84347 100644
@@ -120,7 +120,7 @@ static u8 __init omap_get_die_rev(void)
 
 void __init omap_check_revision(void)
 {
-	int i;
+	int i, sz;
 	u16 jtag_id;
 	u8 die_rev;
 	u32 omap_id;

@@ -194,11 +194,14 @@ void __init omap_check_revision(void)
 		printk(KERN_INFO "Unknown OMAP cpu type: 0x%02x\n", cpu_type);
 	}
 
-	printk(KERN_INFO "OMAP%04x", omap_revision >> 16);
+	sz = snprintf(system_soc_info, SYSTEM_SOC_INFO_SIZE, "OMAP%04x",
+							 omap_revision >> 16);
 	if ((omap_revision >> 8) & 0xff)
-		printk(KERN_INFO "%x", (omap_revision >> 8) & 0xff);
-	printk(KERN_INFO " revision %i handled as %02xxx id: %08x%08x\n",
-	       die_rev, omap_revision & 0xff, system_serial_low,
-	       system_serial_high);
+		snprintf(system_soc_info + sz, SYSTEM_SOC_INFO_SIZE - sz,
+				"%x", (omap_revision >> 8) & 0xff);
+	pr_info("%s revision %i handled as %02xxx id: %08x%08x\n",
+		system_soc_info, die_rev, omap_revision & 0xff,
+		system_serial_low, system_serial_high);
+
 }
Signature · 1 line
1.7.0.4.361.g8b5fe.dirty
author
#4 lore raw

PATCH RESEND v3 4/4 OMAP3: export chip IDCODE, Production ID and Die ID

In reply to Eduardo Valentin

From: Eduardo Valentin <[email protected]>

This patch exports the OMAP3 IDCODE and Production ID to userspace via /proc/cpuinfo using the system_soc_info.

Die ID is also exported depending on what users pass as kernel parameter. It is same protection mechanism made for x86 product number. So, if user passes "omap3_die_id" parameter, it will append die id code into /proc/cpuinfo as well. A Kconfig option has been added as well, so it can be configurable during compilation time.

This can be used to track down silicon specific issues. The info is exported via /proc/cpuinfo because then it can be possible to include this in corematic dumps.

This is based on Peter De Schrijver patch, which export same info via sysfs.

Signed-off-by Eduardo Valentin <[email protected]>
---
 Documentation/kernel-parameters.txt |    2 +
 arch/arm/mach-omap2/Kconfig         |   10 +++++++++
 arch/arm/mach-omap2/id.c            |   37 +++++++++++++++++++++++++++++++++-
 3 files changed, 47 insertions(+), 2 deletions(-)
Documentation/kernel-parameters.txt +2 -0
index 839b21b..6a84e6c 100644
@@ -1809,6 +1809,8 @@ and is between 256 and 4096 characters. It is defined in the file
 			waiting for the ACK, so if this is set too high
 			interrupts *may* be lost!
 
+	omap3_die_id	[OMAP] Append DIE ID info under /proc/cpuinfo
+
 	omap_mux=	[OMAP] Override bootloader pin multiplexing.
 			Format: <mux_mode0.mode_name=value>...
 			For example, to override I2C bus2:
arch/arm/mach-omap2/Kconfig +10 -0
index 2455dcc..462c23a 100644
@@ -169,3 +169,13 @@ config OMAP3_SDRC_AC_TIMING
 	  wish to say no.  Selecting yes without understanding what is
 	  going on could result in system crashes;
 
+config OMAP3_EXPORT_DIE_ID
+	bool "Export DIE ID code under /proc/cpuinfo"
+	depends on ARCH_OMAP3
+	default n
+	help
+	  Say Y here if you need DIE ID code to be exported via /proc/cpuinfo
+	  in production systems. You will need also to explicitly flag it by
+	  appending the "omap3_die_id" parameter to your boot command line.
+
+
arch/arm/mach-omap2/id.c +35 -2
index 4702ffe..6fe554b 100644
@@ -76,6 +76,10 @@ EXPORT_SYMBOL(omap_type);
 /*----------------------------------------------------------------------------*/
 
 #define OMAP_TAP_IDCODE		0x0204
+#define OMAP_TAP_PROD_ID_0      0x0208
+#define OMAP_TAP_PROD_ID_1      0x020c
+#define OMAP_TAP_PROD_ID_2      0x0210
+#define OMAP_TAP_PROD_ID_3      0x0214
 #define OMAP_TAP_DIE_ID_0	0x0218
 #define OMAP_TAP_DIE_ID_1	0x021C
 #define OMAP_TAP_DIE_ID_2	0x0220

@@ -303,6 +307,7 @@ void __init omap4_check_revision(void)
 
 void __init omap3_cpuinfo(void)
 {
+	int sz;
 	u8 rev = GET_OMAP_REVISION();
 	char cpu_name[16], cpu_rev[16];
 

@@ -360,8 +365,8 @@ void __init omap3_cpuinfo(void)
 	}
 
 	/* Print verbose information */
-	snprintf(system_soc_info, SYSTEM_SOC_INFO_SIZE, "%s ES%s", cpu_name,
-			cpu_rev);
+	sz = snprintf(system_soc_info, SYSTEM_SOC_INFO_SIZE, "%s ES%s",
+			cpu_name, cpu_rev);
 	pr_info("%s (", system_soc_info);
 
 	OMAP3_SHOW_FEATURE(l2cache);

@@ -372,7 +377,35 @@ void __init omap3_cpuinfo(void)
 	OMAP3_SHOW_FEATURE(192mhz_clk);
 
 	printk(")\n");
+
+	/* Append OMAP3 IDCODE and Production ID to system_soc_info */
+	snprintf(system_soc_info + sz, SYSTEM_SOC_INFO_SIZE - sz,
+			"\n\tIDCODE\t: %08x\n\tPr. ID\t: %08x %08x %08x %08x",
+			read_tap_reg(OMAP_TAP_IDCODE),
+			read_tap_reg(OMAP_TAP_PROD_ID_0),
+			read_tap_reg(OMAP_TAP_PROD_ID_1),
+			read_tap_reg(OMAP_TAP_PROD_ID_2),
+			read_tap_reg(OMAP_TAP_PROD_ID_3));
+
+}
+
+#ifdef CONFIG_OMAP3_EXPORT_DIE_ID
+static int __init omap3_die_id_setup(char *s)
+{
+	int sz;
+
+	sz = strlen(system_soc_info);
+	snprintf(system_soc_info + sz, SYSTEM_SOC_INFO_SIZE - sz,
+			"\n\tDie ID\t: %08x %08x %08x %08x",
+			read_tap_reg(OMAP_TAP_DIE_ID_0),
+			read_tap_reg(OMAP_TAP_DIE_ID_1),
+			read_tap_reg(OMAP_TAP_DIE_ID_2),
+			read_tap_reg(OMAP_TAP_DIE_ID_3));
+
+	return 1;
 }
+__setup("omap3_die_id", omap3_die_id_setup);
+#endif
 
 /*
  * Try to detect the exact revision of the omap we're running on
Signature · 1 line
1.7.0.4.361.g8b5fe.dirty
#5 lore raw

v3 cover Adding soc related info into /proc/cpuinfo

In reply to Eduardo Valentin
Quoted 2 lines · Eduardo Valentin

I'm resending this series with minor change in the subject prefix of patch 4/4. Removed the PM: prefix. Everything else is same. So, I kept the v3 prefix as well.

I'd much prefer to see a /proc/socinfo rather than extending /proc/cpuinfo to further include stuff which isn't CPU related.

author
#6 lore raw

v3 cover Adding soc related info into /proc/cpuinfo

Hello,

So, back to this track.

Quoted 6 lines · Russell King
Quoted 2 lines · Eduardo Valentin

I'm resending this series with minor change in the subject prefix of patch 4/4. Removed the PM: prefix. Everything else is same. So, I kept the v3 prefix as well.

I'd much prefer to see a /proc/socinfo rather than extending /proc/cpuinfo to further include stuff which isn't CPU related.

Right. I agree that this is not CPU stuff. But I was trying to do not add new interfaces to export those bits.

But the socinfo is actually a nice idea. I will resend this series, but now it will create the socinfo and use it.

As it adds new stuff under /proc, I'll also cc lkml.

Thanks for the suggestion.

BR,

--
Eduardo Valentin