RE: [SPAM] RE: Bug in atomic_sub_if_positive

["Morten Larsen" <mlarsen@broadcom.com>]

Thanks for the reply. My main point is that if the sc instruction fails 
(returns zero) then we need to start over (with another ll instruction.) It 
appears that the current code does not do this correctly. If you have a better 
suggestion for a patch, that's fine with me. Below is the code from the 
2.6.20.21 kernel, which is (also) working for me.

=:-) Morten


                 __asm__ __volatile__(
                 "       .set    mips3                                   \n"
                 "1:     ll      %1, %2          # atomic_sub_if_positive\n"
                 "       subu    %0, %1, %3                              \n"
                 "       bltz    %0, 1f                                  \n"
                 "       sc      %0, %2                                  \n"
                 "       .set    noreorder                               \n"
                 "       beqz    %0, 1b                                  \n"
                 "        subu   %0, %1, %3                              \n"
                 "       .set    reorder                                 \n"
                 "1:                                                     \n"
                 "       .set    mips0                                   \n"
                 : "=&r" (result), "=&r" (temp), "=m" (v->counter)
                 : "Ir" (i), "m" (v->counter)
                 : "memory");



________________________________

        From: Markus Gothe [mailto:markus.gothe@27m.se] 
        Sent: Tuesday, July 01, 2008 5:29 PM
        To: Morten Larsen
        Cc: linux-mips@linux-mips.org
        Subject: Re: [SPAM] RE: Bug in atomic_sub_if_positive
        
        
        NACK. 

        You must realize that 1b stands for 'label 1, backwards', so correctly 
it would be '2: b 1f'... Which is a kind off inconsequent numbering in this 
case.

        //Markus

        On 2 Jul 2008, at 02:12, Morten Larsen wrote:



                        As far as I can tell the branch optimization fixes in 
2.6.21 introduced
                        

                        a bug in atomic_sub_if_positive that causes it to 
return even when the
                        

                        sc instruction fails. The result is that e.g. 
down_trylock becomes
                        

                        unreliable as the semaphore counter is not always 
decremented.
                        


                Previous patch was garbled by Outlook - this one should be 
clean:
                
                --- a/include/asm-mips/atomic.h 2008-06-25 22:38:43.159739000 
-0700
                +++ b/include/asm-mips/atomic.h 2008-06-25 22:39:07.552065000 
-0700
                @@ -292,10 +292,10 @@ static __inline__ int atomic_sub_if_posi
                " beqz %0, 2f \n"
                " subu %0, %1, %3 \n"
                " .set reorder \n"
                - "1: \n"
                " .subsection 2 \n"
                "2: b 1b \n"
                " .previous \n"
                + "1: \n"
                " .set mips0 \n"
                : "=&r" (result), "=&r" (temp), "=m" (v->counter)
                : "Ir" (i), "m" (v->counter)
                @@ -682,10 +682,10 @@ static __inline__ long atomic64_sub_if_p
                " beqz %0, 2f \n"
                " dsubu %0, %1, %3 \n"
                " .set reorder \n"
                - "1: \n"
                " .subsection 2 \n"
                "2: b 1b \n"
                " .previous \n"
                + "1: \n"
                " .set mips0 \n"
                : "=&r" (result), "=&r" (temp), "=m" (v->counter)
                : "Ir" (i), "m" (v->counter)
                


                        _______________________________________

        Mr Markus Gothe
        Software Engineer

        Phone: +46 (0)13 21 81 20 (ext. 1046)
        Fax: +46 (0)13 21 21 15
        Mobile: +46 (0)70 348 44 35
        Diskettgatan 11, SE-583 35 Linköping, Sweden
        www.27m.com