Skip to contentSkip to conversation

lib/vsprintf.c: add %paD format specifier for dma_addr_t types

PATCH In review user.email 5 participants 10 messages Nov 28, 2013 – Dec 3, 2013

Subsystems (3)
lore
Share & tools
lore

user.email

author reviewer
#0 lore raw

PATCH 1/1 lib/vsprintf.c: add %paD format specifier for dma_addr_t types

From: Andy Shevchenko <[email protected]>

Add the %paD format specifier for printing a dma_addr_t type, since the DMA address size on some platforms can vary based on build options, regardless of the native integer type.

---
 Documentation/printk-formats.txt |  7 +++++++
 lib/vsprintf.c                   | 15 ++++++++++++---
 2 files changed, 19 insertions(+), 3 deletions(-)
Documentation/printk-formats.txt +7 -0
index 445ad74..3344ca9 100644
@@ -63,6 +63,13 @@ Physical addresses:
 	resource_size_t) which can vary based on build options, regardless of
 	the width of the CPU data path. Passed by reference.
 
+DMA addresses:
+
+	%paD	0x01234567 or 0x0123456789abcdef
+
+	For printing a dma_addr_t type which can vary based on build options,
+	regardless of the width of the CPU data path. Passed by reference.
+
 Raw buffer as a hex string:
 	%*ph	00 01 02  ...  3f
 	%*phC	00:01:02: ... :3f
lib/vsprintf.c +12 -3
index 10909c5..12ea69d 100644
@@ -1219,6 +1219,7 @@ int kptr_restrict __read_mostly;
  *            The maximum supported length is 64 bytes of the input. Consider
  *            to use print_hex_dump() for the larger input.
  * - 'a' For a phys_addr_t type and its derivative types (passed by reference)
+ * - 'aD' For a dma_addr_t type (passed by reference)
  * - 'd[234]' For a dentry name (optionally 2-4 last components)
  * - 'D[234]' Same as 'd' but for a struct file
  *
@@ -1354,10 +1355,18 @@ char *pointer(const char *fmt, char *buf, char *end, void *ptr,
 		break;
 	case 'a':
 		spec.flags |= SPECIAL | SMALL | ZEROPAD;
-		spec.field_width = sizeof(phys_addr_t) * 2 + 2;
 		spec.base = 16;
-		return number(buf, end,
-			      (unsigned long long) *((phys_addr_t *)ptr), spec);
+		switch (fmt[1]) {
+		case 'D':
+			spec.field_width = sizeof(dma_addr_t) * 2 + 2;
+			return number(buf, end,
+				(unsigned long long)*((dma_addr_t *)ptr), spec);
+		default:
+			spec.field_width = sizeof(phys_addr_t) * 2 + 2;
+			return number(buf, end,
+				(unsigned long long)*((phys_addr_t *)ptr), spec);
+		}
+		break;
 	case 'd':
 		return dentry_name(buf, end, ptr, spec, fmt);
 	case 'D':
Signature · 1 line
1.8.4.4
#1 lore raw
Quoted 5 lines · user.email

From: Andy Shevchenko <[email protected]>

Add the %paD format specifier for printing a dma_addr_t type, since the DMA address size on some platforms can vary based on build options, regardless of the native integer type.

OK, but I prefer this one because upper case is
sometimes used to specify upper case only output.
ala %x %X and %pUl %pUL.

https://lkml.org/lkml/2013/9/12/483

author reviewer
#2 lore raw
Quoted 10 lines · Joe Perches
Quoted 5 lines · user.email

From: Andy Shevchenko <[email protected]>

Add the %paD format specifier for printing a dma_addr_t type, since the DMA address size on some platforms can vary based on build options, regardless of the native integer type.

OK, but I prefer this one because upper case is
sometimes used to specify upper case only output.
ala %x %X and %pUl %pUL.

No objections.

Quoted 1 line

https://lkml.org/lkml/2013/9/12/483

Hmm... Still not in kernel. Do you know reason why it so?

Signature · 2 lines
Andy Shevchenko <[email protected]>
Intel Finland Oy
#3 lore raw
Quoted 12 lines · Andy Shevchenko
Quoted 10 lines · Joe Perches
Quoted 5 lines · user.email

From: Andy Shevchenko <[email protected]>

Add the %paD format specifier for printing a dma_addr_t type, since the DMA address size on some platforms can vary based on build options, regardless of the native integer type.

OK, but I prefer this one because upper case is
sometimes used to specify upper case only output.
ala %x %X and %pUl %pUL.

Quoted 5 lines

No objections.

Quoted 1 line

https://lkml.org/lkml/2013/9/12/483

Hmm... Still not in kernel. Do you know reason why it so?

It was just for discussion and I didn't sign or push it to anyone else.

What's slightly funny is how far back dma_addr_t printk discussions go.

http://lkml.indiana.edu/hypermail/linux/kernel/0202.1/1221.html

Anyway, if you think you're going to do a lot of transforms of printk("%x", dma_addr_t) with or without (unsigned long long) or (u64) casts, I'll submit a signed patch to Andrew.

Let me know.

In an aside for Julia Lawall (added to cc's), I tried to do a spatch grep for dma_addr_t (and phys_addr_t) types that were cast to (unsigned long) or (u64) but I was unsuccessful. Is there something else I need to do?

$ cat dma_addr_t.cocci
@@
dma_addr_t foo;
@@

* (unsigned long long)foo;

$ cat t.c
#include <linux/types.h>
#include <linux/printk.h>

int foo(dma_addr_t a)

{
	printk("test: %llx\n", (unsigned long long)a);
}
$ spatch --all-includes --local-includes  -I include/ --sp-file dma_addr_t.cocci t.c
init_defs_builtins: /usr/local/share/coccinelle/standard.h
HANDLING: t.c
$ 
#4 lore raw
Quoted 39 lines · Joe Perches
Quoted 12 lines · Andy Shevchenko
Quoted 10 lines · Joe Perches
Quoted 5 lines · user.email

From: Andy Shevchenko <[email protected]>

Add the %paD format specifier for printing a dma_addr_t type, since the DMA address size on some platforms can vary based on build options, regardless of the native integer type.

OK, but I prefer this one because upper case is
sometimes used to specify upper case only output.
ala %x %X and %pUl %pUL.

Quoted 5 lines

No objections.

Quoted 1 line

https://lkml.org/lkml/2013/9/12/483

Hmm... Still not in kernel. Do you know reason why it so?

It was just for discussion and I didn't sign or push it to anyone else.

What's slightly funny is how far back dma_addr_t printk discussions go.

http://lkml.indiana.edu/hypermail/linux/kernel/0202.1/1221.html

Anyway, if you think you're going to do a lot of transforms of printk("%x", dma_addr_t) with or without (unsigned long long) or (u64) casts, I'll submit a signed patch to Andrew.

Let me know.

In an aside for Julia Lawall (added to cc's), I tried to do a spatch grep for dma_addr_t (and phys_addr_t) types that were cast to (unsigned long) or (u64) but I was unsuccessful. Is there something else I need to do?

Quoted 6 lines
$ cat dma_addr_t.cocci
@@
dma_addr_t foo;
@@

* (unsigned long long)foo;

Remove the semicolon, otherwise it will only match statements which are nothing but the cast.

Quoted 20 lines
$ cat t.c
#include <linux/types.h>
#include <linux/printk.h>

int foo(dma_addr_t a)

{
	printk("test: %llx\n", (unsigned long long)a);
}
$ spatch --all-includes --local-includes  -I include/ --sp-file dma_addr_t.cocci t.c
init_defs_builtins: /usr/local/share/coccinelle/standard.h
HANDLING: t.c
$

--
To unsubscribe from this list: send the line "unsubscribe dmaengine" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html

#5 lore raw
Quoted 1 line · Lars-Peter Clausen

On 11/29/2013 11:10 PM, Joe Perches wrote:

[]

Quoted 9 lines
Quoted 6 lines
$ cat dma_addr_t.cocci
@@
dma_addr_t foo;
@@

* (unsigned long long)foo;

Remove the semicolon, otherwise it will only match statements which are nothing but the cast.

Duh, thanks.

#6 lore raw
In reply to Joe Perches

Hi Joe,

Quoted 26 lines · On Friday 29 November 2013 14:10:40 Joe Perches wrote:
Quoted 13 lines · Andy Shevchenko
Quoted 11 lines · Joe Perches
Quoted 6 lines · user.email

From: Andy Shevchenko <[email protected]>

Add the %paD format specifier for printing a dma_addr_t type, since the
DMA address size on some platforms can vary based on build options, regardless of the native integer type.

OK, but I prefer this one because upper case is
sometimes used to specify upper case only output.
ala %x %X and %pUl %pUL.

Quoted 5 lines

No objections.

Quoted 1 line

https://lkml.org/lkml/2013/9/12/483

Hmm... Still not in kernel. Do you know reason why it so?

It was just for discussion and I didn't sign or push it to anyone else.

What's slightly funny is how far back dma_addr_t printk discussions go.

Quoted 8 lines

http://lkml.indiana.edu/hypermail/linux/kernel/0202.1/1221.html

Anyway, if you think you're going to do a lot of transforms of printk("%x", dma_addr_t) with or without (unsigned long long) or (u64) casts, I'll submit a signed patch to Andrew.

Let me know.

Just FYI, I've recently submitted a couple of patches that cast dma_addr_t for printk purpose, which resulted in Andy proposing a printk format extension instead.

Quoted 26 lines

In an aside for Julia Lawall (added to cc's), I tried to do a spatch grep for dma_addr_t (and phys_addr_t) types that were cast to (unsigned long) or (u64) but I was unsuccessful. Is there something else I need to do?

$ cat dma_addr_t.cocci
@@
dma_addr_t foo;
@@

* (unsigned long long)foo;

$ cat t.c
#include <linux/types.h>
#include <linux/printk.h>

int foo(dma_addr_t a)

{
	printk("test: %llx\n", (unsigned long long)a);
}
$ spatch --all-includes --local-includes  -I include/ --sp-file
dma_addr_t.cocci t.c init_defs_builtins:
/usr/local/share/coccinelle/standard.h
HANDLING: t.c
$
Signature · 3 lines
Regards,

Laurent Pinchart
author reviewer
#7 lore raw
In reply to Joe Perches
Quoted 2 lines · Joe Perches
Quoted 1 line · Andy Shevchenko

On Thu, 2013-11-28 at 10:32 -0800, Joe Perches wrote:

[]

Quoted 11 lines
Quoted 3 lines
Quoted 1 line

https://lkml.org/lkml/2013/9/12/483

Hmm... Still not in kernel. Do you know reason why it so?

It was just for discussion and I didn't sign or push it to anyone else.

What's slightly funny is how far back dma_addr_t printk discussions go.

http://lkml.indiana.edu/hypermail/linux/kernel/0202.1/1221.html

Wow, 2002!
Let's cross over a finish line in 2013.

Quoted 4 lines

Anyway, if you think you're going to do a lot of transforms of printk("%x", dma_addr_t) with or without (unsigned long long) or (u64) casts, I'll submit a signed patch to Andrew.

I think it worse to submit even if we have no more such casting in the future (which is the case I don't believe in).

Signature · 2 lines
Andy Shevchenko <[email protected]>
Intel Finland Oy
#8 lore raw
Quoted 7 lines · Andy Shevchenko
Quoted 4 lines

Anyway, if you think you're going to do a lot of transforms of printk("%x", dma_addr_t) with or without (unsigned long long) or (u64) casts, I'll submit a signed patch to Andrew.

I think it worse to submit even if we have no more such casting in the future (which is the case I don't believe in).

s/worse/worth/

Signature · 13 lines
Andy Shevchenko <[email protected]>
Intel Finland Oy
---------------------------------------------------------------------
Intel Finland Oy
Registered Address: PL 281, 00181 Helsinki 
Business Identity Code: 0357606 - 4 
Domiciled in Helsinki 

This e-mail and any attachments may contain confidential material for
the sole use of the intended recipient(s). Any review or distribution
by others is strictly prohibited. If you are not the intended
recipient, please contact the sender and delete all copies.
���{.n�+�������+%�����ݶ��w��{.n�+����{��G�����{ay�ʇڙ�,j��f���h��������z_�(�階�ݢj"���m������G������?����&���~��iO��z��v�^�m����������?�I�
#9 lore raw

PATCH 1/1 vsprintf: add %pad extension for dma_addr_t use

dma_addr_t's can be either u32 or u64 depending on a CONFIG option.

There are a few hundred dma_addr_t's printed via either cast to unsigned long long, unsigned long or no cast at all.

Add %pad to be able to emit them without the cast.

Update Documentation/printk-formats.txt too.

---
 Documentation/printk-formats.txt | 11 +++++++++--
 lib/vsprintf.c                   | 33 +++++++++++++++++++++++++++------
 2 files changed, 36 insertions(+), 8 deletions(-)
Documentation/printk-formats.txt +9 -2
index 445ad74..6f4eb32 100644
@@ -55,14 +55,21 @@ Struct Resources:
 	For printing struct resources. The 'R' and 'r' specifiers result in a
 	printed resource with ('R') or without ('r') a decoded flags member.
 
-Physical addresses:
+Physical addresses types phys_addr_t:
 
-	%pa	0x01234567 or 0x0123456789abcdef
+	%pa[p]	0x01234567 or 0x0123456789abcdef
 
 	For printing a phys_addr_t type (and its derivatives, such as
 	resource_size_t) which can vary based on build options, regardless of
 	the width of the CPU data path. Passed by reference.
 
+DMA addresses types dma_addr_t:
+
+	%pad	0x01234567 or 0x0123456789abcdef
+
+	For printing a dma_addr_t type which can vary based on build options,
+	regardless of the width of the CPU data path. Passed by reference.
+
 Raw buffer as a hex string:
 	%*ph	00 01 02  ...  3f
 	%*phC	00:01:02: ... :3f
lib/vsprintf.c +27 -6
index 10909c5..185b6d3 100644
@@ -1155,6 +1155,30 @@ char *netdev_feature_string(char *buf, char *end, const u8 *addr,
 	return number(buf, end, *(const netdev_features_t *)addr, spec);
 }
 
+static noinline_for_stack
+char *address_val(char *buf, char *end, const void *addr,
+		  struct printf_spec spec, const char *fmt)
+{
+	unsigned long long num;
+
+	spec.flags |= SPECIAL | SMALL | ZEROPAD;
+	spec.base = 16;
+
+	switch (fmt[1]) {
+	case 'd':
+		num = *(const dma_addr_t *)addr;
+		spec.field_width = sizeof(dma_addr_t) * 2 + 2;
+		break;
+	case 'p':
+	default:
+		num = *(const phys_addr_t *)addr;
+		spec.field_width = sizeof(phys_addr_t) * 2 + 2;
+		break;
+	}
+
+	return number(buf, end, num, spec);
+}
+
 int kptr_restrict __read_mostly;
 
 /*
@@ -1218,7 +1242,8 @@ int kptr_restrict __read_mostly;
  *              N no separator
  *            The maximum supported length is 64 bytes of the input. Consider
  *            to use print_hex_dump() for the larger input.
- * - 'a' For a phys_addr_t type and its derivative types (passed by reference)
+ * - 'a[pd]' For address types [p] phys_addr_t, [d] dma_addr_t and derivatives
+ *           (default assumed to be phys_addr_t, passed by reference)
  * - 'd[234]' For a dentry name (optionally 2-4 last components)
  * - 'D[234]' Same as 'd' but for a struct file
  *
@@ -1353,11 +1378,7 @@ char *pointer(const char *fmt, char *buf, char *end, void *ptr,
 		}
 		break;
 	case 'a':
-		spec.flags |= SPECIAL | SMALL | ZEROPAD;
-		spec.field_width = sizeof(phys_addr_t) * 2 + 2;
-		spec.base = 16;
-		return number(buf, end,
-			      (unsigned long long) *((phys_addr_t *)ptr), spec);
+		return address_val(buf, end, ptr, spec, fmt);
 	case 'd':
 		return dentry_name(buf, end, ptr, spec, fmt);
 	case 'D':