diff mbox

lib: add weak __clzsi2()/__clzdi2() functions

Message ID 1367214005-14302-1-git-send-email-chanho.min@lge.com
State Not Applicable
Delegated to: David Miller
Headers show

Commit Message

Chanho Min April 29, 2013, 5:40 a.m. UTC
Some architectures need __clzsi2() or __clzdi2() for __builtin_clz and
It causes build failure. They can be implemented using the fls() and
overridden by linking arch-specific versions may not be implemented yet.

Reference: https://lkml.org/lkml/2013/4/18/603
Reported-by: Geert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: Chanho Min <chanho.min@lge.com>
---
 lib/Makefile |    2 +-
 lib/clz.c    |   36 ++++++++++++++++++++++++++++++++++++
 2 files changed, 37 insertions(+), 1 deletion(-)
 create mode 100644 lib/clz.c

Comments

Geert Uytterhoeven April 29, 2013, 7:20 a.m. UTC | #1
On Mon, Apr 29, 2013 at 7:40 AM, Chanho Min <chanho.min@lge.com> wrote:
> Some architectures need __clzsi2() or __clzdi2() for __builtin_clz and
> It causes build failure. They can be implemented using the fls() and
> overridden by linking arch-specific versions may not be implemented yet.
>
> Reference: https://lkml.org/lkml/2013/4/18/603
> Reported-by: Geert Uytterhoeven <geert@linux-m68k.org>
> Signed-off-by: Chanho Min <chanho.min@lge.com>

Thanks, this fixes the link error on m68k, so

Acked-by: Geert Uytterhoeven <geert@linux-m68k.org>

Gr{oetje,eeting}s,

                        Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds
--
To unsubscribe from this list: send the line "unsubscribe sparclinux" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Andrew Morton April 30, 2013, 7:47 p.m. UTC | #2
On Mon, 29 Apr 2013 14:40:05 +0900 Chanho Min <chanho.min@lge.com> wrote:

> Some architectures need __clzsi2() or __clzdi2() for __builtin_clz and
> It causes build failure. They can be implemented using the fls() and
> overridden by linking arch-specific versions may not be implemented yet.
> 
> Reference: https://lkml.org/lkml/2013/4/18/603
> Reported-by: Geert Uytterhoeven <geert@linux-m68k.org>
> Signed-off-by: Chanho Min <chanho.min@lge.com>
>
> ...
>
> --- a/lib/Makefile
> +++ b/lib/Makefile
> @@ -23,7 +23,7 @@ lib-y	+= kobject.o klist.o
>  
>  obj-y += bcd.o div64.o sort.o parser.o halfmd4.o debug_locks.o random32.o \
>  	 bust_spinlocks.o hexdump.o kasprintf.o bitmap.o scatterlist.o \
> -	 gcd.o lcm.o list_sort.o uuid.o flex_array.o \
> +	 gcd.o lcm.o list_sort.o uuid.o flex_array.o clz.o\
>  	 bsearch.o find_last_bit.o find_next_bit.o llist.o memweight.o kfifo.o
>  obj-y += string_helpers.o
>  obj-$(CONFIG_TEST_STRING_HELPERS) += test-string_helpers.o
> diff --git a/lib/clz.c b/lib/clz.c
> new file mode 100644
> index 0000000..5222b2d
> --- /dev/null
> +++ b/lib/clz.c
> @@ -0,0 +1,36 @@
> +/*
> + * lib/clz.c
> + *
> + * Copyright (C) 2013 Chanho Min <chanho.min@lge.com>
> + *
> + * This program is free software; you can redistribute it and/or modify
> + * it under the terms of the GNU General Public License version 2 as
> + * published by the Free Software Foundation.
> + *
> + * __clzsi2/_clzdi2() can be overridden by linking arch-specific versions.
> + */
> +
> +#include <linux/export.h>
> +#include <linux/kernel.h>
> +
> +int __weak __clzsi2(int val)
> +{
> +	return BITS_PER_LONG - fls(val);
> +}
> +EXPORT_SYMBOL(__clzsi2);
> +
> +#if BITS_PER_LONG == 32
> +int __weak __clzdi2(long val)
> +{
> +	return BITS_PER_LONG - fls((int)val);
> +}
> +EXPORT_SYMBOL(__clzdi2);
> +#elif BITS_PER_LONG == 64
> +int __weak __clzdi2i(long val)

This is a typo, surely?  I'll switch it to __clzdi2.

> +{
> +	return BITS_PER_LONG - fls64((u64)val);
> +}
> +EXPORT_SYMBOL(__clzdi2);
> +#else
> +#error BITS_PER_LONG not 32 or 64
> +#endif

--
To unsubscribe from this list: send the line "unsubscribe sparclinux" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Andrew Morton April 30, 2013, 9:46 p.m. UTC | #3
On Mon, 29 Apr 2013 09:20:25 +0200 Geert Uytterhoeven <geert@linux-m68k.org> wrote:

> On Mon, Apr 29, 2013 at 7:40 AM, Chanho Min <chanho.min@lge.com> wrote:
> > Some architectures need __clzsi2() or __clzdi2() for __builtin_clz and
> > It causes build failure. They can be implemented using the fls() and
> > overridden by linking arch-specific versions may not be implemented yet.
> >
> > Reference: https://lkml.org/lkml/2013/4/18/603
> > Reported-by: Geert Uytterhoeven <geert@linux-m68k.org>
> > Signed-off-by: Chanho Min <chanho.min@lge.com>
> 
> Thanks, this fixes the link error on m68k, so
> 
> Acked-by: Geert Uytterhoeven <geert@linux-m68k.org>

It doesn't work for alpha (gcc-4.4.4):

ERROR: "__ctzdi2" [lib/lz4/lz4hc_compress.ko] undefined!
ERROR: "__ctzdi2" [lib/lz4/lz4_compress.ko] undefined!

akpm3:/usr/src/25> nm vmlinux|grep ctzd
akpm3:/usr/src/25> 

I assume the linker threw the code away.


How's about we stop using __builtin_ctz[ll]() altogether?  Use N-fls()?

btw, this:

int __weak __clzsi2(int val)
{
       return BITS_PER_LONG - fls(val);
}

looks wrong.  It will give a different result for 32- and 64-bit.
--
To unsubscribe from this list: send the line "unsubscribe sparclinux" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Geert Uytterhoeven May 1, 2013, 7:35 a.m. UTC | #4
On Tue, Apr 30, 2013 at 11:46 PM, Andrew Morton
<akpm@linux-foundation.org> wrote:
> On Mon, 29 Apr 2013 09:20:25 +0200 Geert Uytterhoeven <geert@linux-m68k.org> wrote:
>> On Mon, Apr 29, 2013 at 7:40 AM, Chanho Min <chanho.min@lge.com> wrote:
>> > Some architectures need __clzsi2() or __clzdi2() for __builtin_clz and
>> > It causes build failure. They can be implemented using the fls() and
>> > overridden by linking arch-specific versions may not be implemented yet.
>> >
>> > Reference: https://lkml.org/lkml/2013/4/18/603
>> > Reported-by: Geert Uytterhoeven <geert@linux-m68k.org>
>> > Signed-off-by: Chanho Min <chanho.min@lge.com>
>>
>> Thanks, this fixes the link error on m68k, so
>>
>> Acked-by: Geert Uytterhoeven <geert@linux-m68k.org>
>
> It doesn't work for alpha (gcc-4.4.4):
>
> ERROR: "__ctzdi2" [lib/lz4/lz4hc_compress.ko] undefined!
> ERROR: "__ctzdi2" [lib/lz4/lz4_compress.ko] undefined!
>
> akpm3:/usr/src/25> nm vmlinux|grep ctzd
> akpm3:/usr/src/25>
>
> I assume the linker threw the code away.

No, the link errors are for __ctzdi2, not __clzdi2.
__ctzdi2 can be implemented using __ffs().

> How's about we stop using __builtin_ctz[ll]() altogether?  Use N-fls()?

That was my first suggestion ;-)

> btw, this:
>
> int __weak __clzsi2(int val)
> {
>        return BITS_PER_LONG - fls(val);
> }
>
> looks wrong.  It will give a different result for 32- and 64-bit.

Indeed. It should be hardcoded 32.

Gr{oetje,eeting}s,

                        Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds
--
To unsubscribe from this list: send the line "unsubscribe sparclinux" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Chanho Min May 2, 2013, 12:58 a.m. UTC | #5
>No, the link errors are for __ctzdi2, not __clzdi2.
>__ctzdi2 can be implemented using __ffs().
>> How's about we stop using __builtin_ctz[ll]() altogether?  Use N-fls()?
>
>That was my first suggestion ;-)
>
Do we remove existing __builtin_ct[l]z?
Some architecture's __builtin_clz will be more optimized than N-fls
IMHO, This is not only about __builtin- but also supporting another
libgcc functions in kernel.
Some future user may need __clz[t]s[d]i2 symbol also.

>> btw, this:
>>
>> int __weak __clzsi2(int val)
>> {
>>        return BITS_PER_LONG - fls(val);
>> }
>>
>> looks wrong.  It will give a different result for 32- and 64-bit.
>
>Indeed. It should be hardcoded 32.
>
I will re-work this for 64-bit, and ctz also.

Thanks
Chanho Min

--
To unsubscribe from this list: send the line "unsubscribe sparclinux" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
diff mbox

Patch

diff --git a/lib/Makefile b/lib/Makefile
index af79e8c..e17b3ee 100644
--- a/lib/Makefile
+++ b/lib/Makefile
@@ -23,7 +23,7 @@  lib-y	+= kobject.o klist.o
 
 obj-y += bcd.o div64.o sort.o parser.o halfmd4.o debug_locks.o random32.o \
 	 bust_spinlocks.o hexdump.o kasprintf.o bitmap.o scatterlist.o \
-	 gcd.o lcm.o list_sort.o uuid.o flex_array.o \
+	 gcd.o lcm.o list_sort.o uuid.o flex_array.o clz.o\
 	 bsearch.o find_last_bit.o find_next_bit.o llist.o memweight.o kfifo.o
 obj-y += string_helpers.o
 obj-$(CONFIG_TEST_STRING_HELPERS) += test-string_helpers.o
diff --git a/lib/clz.c b/lib/clz.c
new file mode 100644
index 0000000..5222b2d
--- /dev/null
+++ b/lib/clz.c
@@ -0,0 +1,36 @@ 
+/*
+ * lib/clz.c
+ *
+ * Copyright (C) 2013 Chanho Min <chanho.min@lge.com>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * __clzsi2/_clzdi2() can be overridden by linking arch-specific versions.
+ */
+
+#include <linux/export.h>
+#include <linux/kernel.h>
+
+int __weak __clzsi2(int val)
+{
+	return BITS_PER_LONG - fls(val);
+}
+EXPORT_SYMBOL(__clzsi2);
+
+#if BITS_PER_LONG == 32
+int __weak __clzdi2(long val)
+{
+	return BITS_PER_LONG - fls((int)val);
+}
+EXPORT_SYMBOL(__clzdi2);
+#elif BITS_PER_LONG == 64
+int __weak __clzdi2i(long val)
+{
+	return BITS_PER_LONG - fls64((u64)val);
+}
+EXPORT_SYMBOL(__clzdi2);
+#else
+#error BITS_PER_LONG not 32 or 64
+#endif