[Top] [All Lists]

Re: bitops.h ext2_ ops save/restore flags

Subject: Re: bitops.h ext2_ ops save/restore flags
From: Florian Lohoff <>
Date: Tue, 19 Jun 2001 16:36:51 +0200
In-reply-to: <>; from on Tue, Jun 19, 2001 at 04:23:02PM +0200
Organization: rfc822 - pure communication
References: <>
User-agent: Mutt/1.3.17i
On Tue, Jun 19, 2001 at 04:23:02PM +0200, Florian Lohoff wrote:
> Hi,
> i am just staring at the bitops as it seems they are the culprit for
> the non working ext2 currently - While staring and comparing i found
> that the mips ext2_set_bit and co do an save_and_cli/restore_flags
> which is very obscure. I cant find any similar in other archs - I guess
> the above is due to a possible "atomic" requirement which ext2 doesnt
> seem to have (upper layer locking ?)

Another one - This only fixes the big endian case - the little
endian case still has the penalty of cli/restore on every bit set
as the ext2 macros get defined to the "atomic" ops defined there

I propose this patch

Index: include/asm-mips/bitops.h
RCS file: /cvs/linux/include/asm-mips/bitops.h,v
retrieving revision 1.16
diff -u -r1.16 bitops.h
--- include/asm-mips/bitops.h   2001/06/14 05:33:12     1.16
+++ include/asm-mips/bitops.h   2001/06/19 14:32:51
@@ -887,8 +883,8 @@
 #else /* !(__MIPSEB__) */
 /* Native ext2 byte ordering, just collapse using defines. */
-#define ext2_set_bit(nr, addr) test_and_set_bit((nr), (addr))
-#define ext2_clear_bit(nr, addr) test_and_clear_bit((nr), (addr))
+#define ext2_set_bit(nr, addr) __test_and_set_bit((nr), (addr))
+#define ext2_clear_bit(nr, addr) __test_and_clear_bit((nr), (addr))
 #define ext2_test_bit(nr, addr) test_bit((nr), (addr))
 #define ext2_find_first_zero_bit(addr, size) find_first_zero_bit((addr), 
 #define ext2_find_next_zero_bit(addr, size, offset) \

Comments ?

Florian Lohoff                     +49-5201-669912
     Why is it called "common sense" when nobody seems to have any?

<Prev in Thread] Current Thread [Next in Thread>