Skip to contentSkip to conversation

x86/bootflag: Change some static functions to bool

PATCH Applied Uros Bizjak 5 participants 15 messages Jan 29, 2025 – Feb 26, 2025 x86

lore
Share & tools
lore
#0 lore raw

PATCH 1/2 x86/bootflag: Change some static functions to bool

The return values of some functions are of boolean type. Change the type of these function to bool and adjust their return values.

No functional change intended.

Cc Thomas Gleixner <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Borislav Petkov <[email protected]>
Cc Dave Hansen <[email protected]>
Cc "H. Peter Anvin" <[email protected]>
---
 arch/x86/kernel/bootflag.c | 12 ++++++------
 1 file changed, 6 insertions(+), 6 deletions(-)
arch/x86/kernel/bootflag.c +6 -6
index 3fed7ae58b60..4d89a2d80d0f 100644
@@ -20,7 +20,7 @@
 
 int sbf_port __initdata = -1;	/* set via acpi_boot_init() */
 
-static int __init parity(u8 v)
+static bool __init parity(u8 v)
 {
 	int x = 0;
 	int i;
@@ -30,7 +30,7 @@ static int __init parity(u8 v)
 		v >>= 1;
 	}
 
-	return x;
+	return !!x;
 }
 
 static void __init sbf_write(u8 v)
@@ -66,14 +66,14 @@ static u8 __init sbf_read(void)
 	return v;
 }
 
-static int __init sbf_value_valid(u8 v)
+static bool __init sbf_value_valid(u8 v)
 {
 	if (v & SBF_RESERVED)		/* Reserved bits */
-		return 0;
+		return false;
 	if (!parity(v))
-		return 0;
+		return false;
 
-	return 1;
+	return true;
 }
 
 static int __init sbf_init(void)
Signature · 1 line
2.42.0
#1 lore raw

PATCH 2/2 x86/bootflag: Use __builtin_parity() when available

Compilers (GCC and clang) provide optimized __builtin_parity() function that returns the parity of X, i.e. the number of 1-bits in X modulo 2.

Use __builtin_parity() built-in function to optimize parity(). This improves generated parity code to just:

  57:	84 db                	test   %bl,%bl
  59:	7a 5c                	jp     b7 <sbf_init+0xa7>

where TEST instruction sets parity flag (PF) in EFLAGS, exercised by the follow-up jump instruction.

Cc Thomas Gleixner <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Borislav Petkov <[email protected]>
Cc Dave Hansen <[email protected]>
Cc "H. Peter Anvin" <[email protected]>
---
 arch/x86/kernel/bootflag.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)
arch/x86/kernel/bootflag.c +4 -1
index 4d89a2d80d0f..d63247d7abd4 100644
@@ -22,6 +22,9 @@ int sbf_port __initdata = -1;	/* set via acpi_boot_init() */
 
 static bool __init parity(u8 v)
 {
+#if __has_builtin(__builtin_parity)
+	int x = __builtin_parity(v);
+#else
 	int x = 0;
 	int i;
 
@@ -29,7 +32,7 @@ static bool __init parity(u8 v)
 		x ^= (v & 1);
 		v >>= 1;
 	}
-
+#endif
 	return !!x;
 }
 
Signature · 1 line
2.42.0
#2 lore raw

Re: [PATCH 2/2] x86/bootflag: Use __builtin_parity() when available

Quoted 4 lines · Uros Bizjak

Compilers (GCC and clang) provide optimized __builtin_parity() function that returns the parity of X, i.e. the number of 1-bits in X modulo 2.

Quoted 8 lines

Use __builtin_parity() built-in function to optimize parity(). This improves generated parity code to just:

  57:   84 db                   test   %bl,%bl
  59:   7a 5c                   jp     b7 <sbf_init+0xa7>

where TEST instruction sets parity flag (PF) in EFLAGS, exercised by the follow-up jump instruction.

Unfortunately, this approach won't fly. As said by Pedro elsewhere:

__builtin functions aren't strictly required to be inlined and can generate a library call if the compiler so chooses (how often this happens depends on the optimization level and architecture). e.g:
https://godbolt.org/z/zKPb9hbaM

Patch is retracted.

Uros.

Quoted 37 lines
Cc Thomas Gleixner <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Borislav Petkov <[email protected]>
Cc Dave Hansen <[email protected]>
Cc "H. Peter Anvin" <[email protected]>
---
 arch/x86/kernel/bootflag.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)
arch/x86/kernel/bootflag.c +4 -2
index 4d89a2d80d0f..d63247d7abd4 100644
@@ -22,6 +22,9 @@ int sbf_port __initdata = -1; /* set via acpi_boot_init() */
 
 static bool __init parity(u8 v)
 {
+#if __has_builtin(__builtin_parity)
+       int x = __builtin_parity(v);
+#else
        int x = 0;
        int i;
 
@@ -29,7 +32,7 @@ static bool __init parity(u8 v)
                x ^= (v & 1);
                v >>= 1;
        }
-
+#endif
        return !!x;
 }
 
--

2.42.0

merge bot Applied
#3 lore raw

[tip: x86/boot] x86/boot: Change some static bootflag functions to bool

In reply to Uros Bizjak
The following commit has been merged into the x86/boot branch of tip:

The following commit has been merged into the x86/boot branch of tip:

Commit-ID:     5bebe2e4fe27bf68b4993d62be2f8bae9e6229d6
Gitweb:        https://git.kernel.org/tip/5bebe2e4fe27bf68b4993d62be2f8bae9e6229d6
Author:        Uros Bizjak <[email protected]>
AuthorDate:    Wed, 29 Jan 2025 16:47:50 +01:00
Committer:     Ingo Molnar <[email protected]>
CommitterDate: Sat, 22 Feb 2025 12:30:59 +01:00

x86/boot: Change some static bootflag functions to bool

The return values of some functions are of boolean type. Change the type of these function to bool and adjust their return values.

No functional change intended.

---
 arch/x86/kernel/bootflag.c | 12 ++++++------
 1 file changed, 6 insertions(+), 6 deletions(-)
arch/x86/kernel/bootflag.c +6 -6
index 3fed7ae..4d89a2d 100644
@@ -20,7 +20,7 @@
 
 int sbf_port __initdata = -1;	/* set via acpi_boot_init() */
 
-static int __init parity(u8 v)
+static bool __init parity(u8 v)
 {
 	int x = 0;
 	int i;
@@ -30,7 +30,7 @@ static int __init parity(u8 v)
 		v >>= 1;
 	}
 
-	return x;
+	return !!x;
 }
 
 static void __init sbf_write(u8 v)
@@ -66,14 +66,14 @@ static u8 __init sbf_read(void)
 	return v;
 }
 
-static int __init sbf_value_valid(u8 v)
+static bool __init sbf_value_valid(u8 v)
 {
 	if (v & SBF_RESERVED)		/* Reserved bits */
-		return 0;
+		return false;
 	if (!parity(v))
-		return 0;
+		return false;
 
-	return 1;
+	return true;
 }
 
 static int __init sbf_init(void)
#4 lore raw
In reply to Uros Bizjak
Quoted 29 lines · Uros Bizjak

The return values of some functions are of boolean type. Change the type of these function to bool and adjust their return values.

No functional change intended.

Cc Thomas Gleixner <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Borislav Petkov <[email protected]>
Cc Dave Hansen <[email protected]>
Cc "H. Peter Anvin" <[email protected]>
---
  arch/x86/kernel/bootflag.c | 12 ++++++------
  1 file changed, 6 insertions(+), 6 deletions(-)
arch/x86/kernel/bootflag.c +1 -1
index 3fed7ae58b60..4d89a2d80d0f 100644
@@ -20,7 +20,7 @@
 
  int sbf_port __initdata = -1;	/* set via acpi_boot_init() */
 
-static int __init parity(u8 v)
+static bool __init parity(u8 v)
  {
  	int x = 0;
  	int i;

arch/x86/kernel/bootflag.c static int __init parity(u8 v) · line 33

Quoted 5 lines
arch/x86/kernel/bootflag.c +1 -1
@@ -30,7 +30,7 @@ static int __init parity(u8 v)
  		v >>= 1;
  	}
 
-	return x;
+	return !!x;

This "!!" is unnecessary and only obfuscates the code, right?

thanks,

Signature · 2 lines
js
suse labs
#5 lore raw
Quoted 34 lines · Jiri Slaby
Quoted 32 lines · Uros Bizjak

The return values of some functions are of boolean type. Change the type of these function to bool and adjust their return values.

No functional change intended.

Cc Thomas Gleixner <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Borislav Petkov <[email protected]>
Cc Dave Hansen <[email protected]>
Cc "H. Peter Anvin" <[email protected]>
---
  arch/x86/kernel/bootflag.c | 12 ++++++------
  1 file changed, 6 insertions(+), 6 deletions(-)
arch/x86/kernel/bootflag.c +1 -1
index 3fed7ae58b60..4d89a2d80d0f 100644
@@ -20,7 +20,7 @@
 
  int sbf_port __initdata = -1;       /* set via acpi_boot_init() */
 
-static int __init parity(u8 v)
+static bool __init parity(u8 v)
  {
      int x = 0;
      int i;
@@ -30,7 +30,7 @@ static int __init parity(u8 v)
              v >>= 1;
      }
 
Quoted 4 lines

arch/x86/kernel/bootflag.c static int __init parity(u8 v) · line 33

Quoted 2 lines
arch/x86/kernel/bootflag.c +1 -1
-     return x;
+     return !!x;

This "!!" is unnecessary and only obfuscates the code, right?

Not really, this idiom is used in place of (x != 0) to change the type to the return type of the function in a pedantic way.

Uros.

#6 lore raw
Quoted 35 lines · Uros Bizjak
Quoted 34 lines · Jiri Slaby
Quoted 32 lines · Uros Bizjak

The return values of some functions are of boolean type. Change the type of these function to bool and adjust their return values.

No functional change intended.

Cc Thomas Gleixner <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Borislav Petkov <[email protected]>
Cc Dave Hansen <[email protected]>
Cc "H. Peter Anvin" <[email protected]>
---
   arch/x86/kernel/bootflag.c | 12 ++++++------
   1 file changed, 6 insertions(+), 6 deletions(-)
arch/x86/kernel/bootflag.c +1 -1
index 3fed7ae58b60..4d89a2d80d0f 100644
@@ -20,7 +20,7 @@
 
   int sbf_port __initdata = -1;       /* set via acpi_boot_init() */
 
-static int __init parity(u8 v)
+static bool __init parity(u8 v)
   {
       int x = 0;
       int i;
@@ -30,7 +30,7 @@ static int __init parity(u8 v)
               v >>= 1;
       }
 
Quoted 7 lines
Quoted 4 lines

arch/x86/kernel/bootflag.c static int __init parity(u8 v) · line 33

Quoted 2 lines
arch/x86/kernel/bootflag.c +1 -1
-     return x;
+     return !!x;

This "!!" is unnecessary and only obfuscates the code, right?

Not really, this idiom is used in place of (x != 0) to change the type to the return type of the function in a pedantic way.

Care to explain what exactly it changes?

Signature · 2 lines
js
suse labs
#7 lore raw
Quoted 42 lines · Jiri Slaby
Quoted 40 lines · Uros Bizjak
Quoted 38 lines · Jiri Slaby
Quoted 29 lines · Uros Bizjak

The return values of some functions are of boolean type. Change the type of these function to bool and adjust their return values.

No functional change intended.

Cc Thomas Gleixner <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Borislav Petkov <[email protected]>
Cc Dave Hansen <[email protected]>
Cc "H. Peter Anvin" <[email protected]>
---
   arch/x86/kernel/bootflag.c | 12 ++++++------
   1 file changed, 6 insertions(+), 6 deletions(-)
arch/x86/kernel/bootflag.c +1 -1
index 3fed7ae58b60..4d89a2d80d0f 100644
@@ -20,7 +20,7 @@
 
   int sbf_port __initdata = -1;       /* set via acpi_boot_init() */
 
-static int __init parity(u8 v)
+static bool __init parity(u8 v)
   {
       int x = 0;
       int i;

arch/x86/kernel/bootflag.c static int __init parity(u8 v) · line 33

Quoted 5 lines
arch/x86/kernel/bootflag.c +1 -1
@@ -30,7 +30,7 @@ static int __init parity(u8 v)
               v >>= 1;
       }
 
-     return x;
+     return !!x;

This "!!" is unnecessary and only obfuscates the code, right?

Quoted 4 lines
Quoted 2 lines

Not really, this idiom is used in place of (x != 0) to change the type to the return type of the function in a pedantic way.

Care to explain what exactly it changes?

The internal compiler representation of the following testcase:

_Bool foo (int x) { return x; }

is:

--cut here--
_Bool foo (int x)
{
_Bool _2;

<bb 2> [local count: 1073741824]:
_2 = x_1(D) != 0;
return _2;
}
--cut here--

For me, !!x in the source means that the change of types was intentional. Surely, the compiler can do it by itself, so at the end of the day, it is just a matter of personal taste.

FYI, the whole function will soon be removed and replaced with a generic parity8() function.

Uros.

#8 lore raw
Quoted 6 lines · Uros Bizjak

The internal compiler representation of the following testcase:

_Bool foo (int x) { return x; }

is:

Quoted 8 lines

--cut here--
_Bool foo (int x)
{
_Bool _2;

<bb 2> [local count: 1073741824]:
_2 = x_1(D) != 0;
return _2;

Yes, exactly as dictated by the C99 standard.

Quoted 6 lines

}
--cut here--

For me, !!x in the source means that the change of types was intentional. Surely, the compiler can do it by itself, so at the end of the day, it is just a matter of personal taste.

I've just learnt, that we even have that in CodingStyle:
===

Quoted 7 lines

17) Using bool
--------------

The Linux kernel bool type is an alias for the C99 _Bool type. bool values can only evaluate to 0 or 1, and implicit or explicit conversion to bool automatically converts the value to true or false. When using bool types the !! construction is not needed, which eliminates a class of bugs.

===

Quoted 2 lines

FYI, the whole function will soon be removed and replaced with a generic parity8() function.

Yeah, your post in that thread brought me here, in fact. Though, I didn't realize it actually removes this very function in full.

thanks,

Signature · 2 lines
js
suse labs
#9 lore raw
Quoted 19 lines · Jiri Slaby
Quoted 6 lines · Uros Bizjak

The internal compiler representation of the following testcase:

_Bool foo (int x) { return x; }

is:

Quoted 8 lines

--cut here--
_Bool foo (int x)
{
_Bool _2;

<bb 2> [local count: 1073741824]:
_2 = x_1(D) != 0;
return _2;

Yes, exactly as dictated by the C99 standard.

Quoted 8 lines
Quoted 6 lines

}
--cut here--

For me, !!x in the source means that the change of types was intentional. Surely, the compiler can do it by itself, so at the end of the day, it is just a matter of personal taste.

I've just learnt, that we even have that in CodingStyle:

Ugh, I didn't know that either... Thanks for the info.

Regarding the issue itself, I can create a follow-up patch that removes !!x, or perhaps we can go all the way and substitute the function with parity8() generic function, available in bitops.h.

Thanks,
Uros.

#10 lore raw
In reply to Jiri Slaby
Quoted 27 lines · Jiri Slaby
Quoted 6 lines · Uros Bizjak

The internal compiler representation of the following testcase:

_Bool foo (int x) { return x; }

is:

Quoted 8 lines

--cut here--
_Bool foo (int x)
{
_Bool _2;

<bb 2> [local count: 1073741824]:
_2 = x_1(D) != 0;
return _2;

Yes, exactly as dictated by the C99 standard.

Quoted 6 lines

}
--cut here--

For me, !!x in the source means that the change of types was intentional. Surely, the compiler can do it by itself, so at the end of the day, it is just a matter of personal taste.

I've just learnt, that we even have that in CodingStyle:
===

Quoted 8 lines
Quoted 7 lines

17) Using bool
--------------

The Linux kernel bool type is an alias for the C99 _Bool type. bool values can only evaluate to 0 or 1, and implicit or explicit conversion to bool automatically converts the value to true or false. When using bool types the !! construction is not needed, which eliminates a class of bugs.

===

This rule doesn't apply here, because the !! operation isn't done on bool types: 'x' in the parity() function is an 'int'...

So this CodingStyle entry is a red herring, and the !! is absolutely used in the kernel as an explicit marker of intentional type conversion to bool.

Thanks,

Ingo

#11 lore raw
Quoted 2 lines · Ingo Molnar

So this CodingStyle entry is a red herring, and the !! is absolutely used in the kernel

Sure, for intended conversion to either 0 or 1.

Quoted 2 lines

as an explicit marker of intentional type conversion to bool.

With this in mind, you would have to write "if (!!x)" everywhere.

I don't want such constructions in code I maintain. (Nor for return values.) But this is not code I maintain (obviously), so your call after all.

thanks,

Signature · 2 lines
js
suse labs
#12 lore raw
Quoted 6 lines · Jiri Slaby
Quoted 2 lines · Ingo Molnar

So this CodingStyle entry is a red herring, and the !! is absolutely used in the kernel

Sure, for intended conversion to either 0 or 1.

Quoted 8 lines
Quoted 2 lines

as an explicit marker of intentional type conversion to bool.

With this in mind, you would have to write "if (!!x)" everywhere.

I don't want such constructions in code I maintain. (Nor for return values.) But this is not code I maintain (obviously), so your call after all.

thanks,

Uh, no, that's not the point.

The point is that:

!!x

... is the same as

x ? true : false

... which if promoted to an integer, intentionally or not, becomes

x ? 1 : 0

#13 lore raw
Quoted 4 lines · H. Peter Anvin
Quoted 3 lines · Jiri Slaby
Quoted 2 lines · Ingo Molnar

So this CodingStyle entry is a red herring, and the !! is absolutely used in the kernel

label1:

Quoted 20 lines
Quoted 5 lines

Sure, for intended conversion to either 0 or 1.

Quoted 8 lines
Quoted 2 lines

as an explicit marker of intentional type conversion to bool.

With this in mind, you would have to write "if (!!x)" everywhere.

I don't want such constructions in code I maintain. (Nor for return values.) But this is not code I maintain (obviously), so your call after all.

thanks,

Uh, no, that's not the point.

The point is that:

!!x

Quoted 7 lines

... is the same as

x ? true : false

... which if promoted to an integer, intentionally or not, becomes

x ? 1 : 0

Which I write at label1 above. Sorry, am I missing something?

"!!x" when implicitly converted to bool is the very same as simple "x".

Signature · 2 lines
js
suse labs
#14 lore raw
In reply to Jiri Slaby
Quoted 10 lines · Jiri Slaby
Quoted 2 lines · Ingo Molnar

So this CodingStyle entry is a red herring, and the !! is absolutely used in the kernel

Sure, for intended conversion to either 0 or 1.

Quoted 2 lines

as an explicit marker of intentional type conversion to bool.

With this in mind, you would have to write "if (!!x)" everywhere.

No, why would I? In a conditional statement any type conversion is for that evaluation alone and any mistakes are limited to that statement.

On a return statement the value continues to live on in the call context and has a far longer lifetime. Marking that the type conversion from int to bool was intentional is prudent, and the use of '!!' is common practice within the kernel:

starship:~/tip> git grep '!!' -- '*.[ch]' | wc -l 10739

Thanks,

Ingo