Hi Ray, On 11/07/18 16:54, Ruiyu Ni wrote: > Today's code generates assembly code as below for > InternalSyncIncrement: > __asm__ __volatile__ ( > "movl $1, %%eax \n\t" > "lock \n\t" > "xadd %%eax, %1 \n\t" > "inc %%eax \n\t" > : "=a" (Result), // %0 > "+m" (*Value) // %1 > : // no inputs that aren't also outputs > : "memory", > "cc" > ); > > 0: 55 pushl %ebp > 1: 89 e5 movl %esp, %ebp > 3: 8b 45 08 movl 8(%ebp), %eax > 6: b8 01 00 00 00 movl $1, %eax > b: f0 lock > c: 0f c1 00 xaddl %eax, _InternalSyncIncrement(%eax) > f: 40 incl %eax > 10: 5d popl %ebp > 11: c3 retl > > Line #3 and Line #6 both use EAX as destination register. > Line #c uses EAX and (EAX). > > The output operand "=a" tells GCC that EAX is used for output. > But GCC only assumes that EAX will be used in the very last > instruction. > > Per GCC document, > "Use the ‘&’ constraint modifier on all output operands that must > not overlap an input. Otherwise, GCC may allocate the output > operand in the same register as an unrelated input operand, on > the assumption that the assembler code consumes its inputs before > producing outputs. This assumption may be false if the assembler > code actually consists of more than one instruction." > > "=&a" should be used to tell GCC not use EAX before the assembly. > > Contributed-under: TianoCore Contribution Agreement 1.1 > Signed-off-by: Ruiyu Ni <[email protected]> > Cc: Liming Gao <[email protected]> > Cc: Michael Kinney <[email protected]> > Cc: Laszlo Ersek <[email protected]> > Cc: Philippe Mathieu-Daudé <[email protected]> > Cc: Ard Biesheuvel <[email protected]> > --- > MdePkg/Library/BaseSynchronizationLib/Ia32/GccInline.c | 4 ++-- > MdePkg/Library/BaseSynchronizationLib/X64/GccInline.c | 4 ++-- > 2 files changed, 4 insertions(+), 4 deletions(-) > > diff --git a/MdePkg/Library/BaseSynchronizationLib/Ia32/GccInline.c > b/MdePkg/Library/BaseSynchronizationLib/Ia32/GccInline.c > index af39bdeb51..760a020a32 100644 > --- a/MdePkg/Library/BaseSynchronizationLib/Ia32/GccInline.c > +++ b/MdePkg/Library/BaseSynchronizationLib/Ia32/GccInline.c > @@ -40,7 +40,7 @@ InternalSyncIncrement ( > "lock \n\t" > "xadd %%eax, %1 \n\t" > "inc %%eax \n\t" > - : "=a" (Result), // %0 > + : "=&a" (Result), // %0 > "+m" (*Value) // %1 > : // no inputs that aren't also outputs > : "memory", > @@ -76,7 +76,7 @@ InternalSyncDecrement ( > "lock \n\t" > "xadd %%eax, %1 \n\t" > "dec %%eax \n\t" > - : "=a" (Result), // %0 > + : "=&a" (Result), // %0 > "+m" (*Value) // %1 > : // no inputs that aren't also outputs > : "memory", > diff --git a/MdePkg/Library/BaseSynchronizationLib/X64/GccInline.c > b/MdePkg/Library/BaseSynchronizationLib/X64/GccInline.c > index edb904c007..767d4626b8 100644 > --- a/MdePkg/Library/BaseSynchronizationLib/X64/GccInline.c > +++ b/MdePkg/Library/BaseSynchronizationLib/X64/GccInline.c > @@ -40,7 +40,7 @@ InternalSyncIncrement ( > "lock \n\t" > "xadd %%eax, %1 \n\t" > "inc %%eax \n\t" > - : "=a" (Result), // %0 > + : "=&a" (Result), // %0 > "+m" (*Value) // %1 > : // no inputs that aren't also outputs > : "memory", > @@ -76,7 +76,7 @@ InternalSyncDecrement ( > "lock \n\t" > "xadd %%eax, %1 \n\t" > "dec %%eax \n\t" > - : "=a" (Result), // %0 > + : "=&a" (Result), // %0 > "+m" (*Value) // %1 > : // no inputs that aren't also outputs > : "memory", >
(1) Before you push the patch, please reinstate the reference to: https://bugzilla.tianocore.org/show_bug.cgi?id=1303 in the commit message. (The BZ reference was part of v1.) (2) Please also append: Suggested-by: Ard Biesheuvel <[email protected]> Fixes: 8a94eb9283fa09a30f5f06f0c12cf0ee4e14fbcf Fixes: 17634d026f968c404b039a8d8431b6389dd396ea (3) Can you please replace the ‘ (U+2018) and ’ (U+2019) characters in the commit message with simple ASCII quotes, such as '&'? Unless there's a strong reason, I prefer to limit commit messages to plain ASCII. (This does not apply to names of people, of course.) Feel free to ignore my request if you disagree. (4) The patch looks good to me. Reviewed-by: Laszlo Ersek <[email protected]> (5) I've also tested the patch, on top of c60d36b4d1ee, both the IA32 and the X64 changes, by performing my usual Linux guest tests described at <https://github.com/tianocore/tianocore.github.io/wiki/Testing-SMM-with-QEMU,-KVM-and-libvirt#tests-to-perform-in-the-installed-guest-fedora-26-guest> with the IA32, IA32X64, and X64 builds of OVMF. Tested-by: Laszlo Ersek <[email protected]> Thanks! Laszlo _______________________________________________ edk2-devel mailing list [email protected] https://lists.01.org/mailman/listinfo/edk2-devel
