[PATCH] arm: prevent inlining in arxescsi.c causing build failures
Arnd Bergmann
arnd at arndb.de
Thu Dec 1 09:06:49 EST 2011
On Thursday 01 December 2011, Uwe Kleine-König wrote:
> On Wed, Nov 30, 2011 at 11:26:51PM -0500, Paul Gortmaker wrote:
> > The following failure is seen in routine coverage builds:
> >
> > CC [M] drivers/scsi/arm/arxescsi.o
> > /tmp/cccEyiO7.s: Assembler messages:
> > /tmp/cccEyiO7.s:334: Error: symbol `.loop_1' is already defined
> > /tmp/cccEyiO7.s:337: Error: symbol `.loop_2' is already defined
> > /tmp/cccEyiO7.s:343: Error: symbol `.loop_3' is already defined
> > /tmp/cccEyiO7.s:365: Error: symbol `.loop_1' is already defined
> > /tmp/cccEyiO7.s:368: Error: symbol `.loop_2' is already defined
> > /tmp/cccEyiO7.s:374: Error: symbol `.loop_3' is already defined
> > make[4]: *** [drivers/scsi/arm/arxescsi.o] Error 1
> >
> > It is caused by multiple inline of arxescsi_pseudo_dma_write()
> > which is responsible for the above labels. Marking the fcn
> > as non-inline fixes the issue.
> >
> > Signed-off-by: Paul Gortmaker <paul.gortmaker at windriver.com>
> Is that a compiler or a code issue? If it's the compiler please fix
> that. If it's the code, then please document why you added the noinline.
The problem is mostly in the code: The symbols are defined in an
inline assembly that is not meant to be instantiated multiple times.
gcc may decide to unroll the loop in which this is done, which causes
the error above. Unrolling the loop is a rather silly thing to do
here, because the code is rather large and does not at all benefit
from unrolling. You can use the patch below to fix the code to still
work if the loop is unrolled, but I also think that marking the
function as uninline is a good idea, in particular because I don't
trust the old assembly to still do the right thing otherwise:
It manually saves and restores the registers on the stack, where
you nowadays would specify specific clobbers for the registers
it uses, and let the compiler take care of it. It also doesn't
contain a memory clobber for the buffer, so keeping the function
out-of-line is probably the safe choice.
> Having said that, "my" compiler compiles drivers/scsi/arm/arxescsi.o
> just fine (using rpc_defconfig on v3.2-rc2).
You should be able to reproduce it by adding -funroll-all-loops to
the CFLAGS. That made the difference for me on gcc-4.6 from Ubuntu.
Arnd
8<-----
PATCH: scsi/arxcescsi: use local assembler symbols
> > /tmp/cccEyiO7.s: Assembler messages:
> > /tmp/cccEyiO7.s:334: Error: symbol `.loop_1' is already defined
> > /tmp/cccEyiO7.s:337: Error: symbol `.loop_2' is already defined
> > /tmp/cccEyiO7.s:343: Error: symbol `.loop_3' is already defined
> > /tmp/cccEyiO7.s:365: Error: symbol `.loop_1' is already defined
> > /tmp/cccEyiO7.s:368: Error: symbol `.loop_2' is already defined
> > /tmp/cccEyiO7.s:374: Error: symbol `.loop_3' is already defined
> > make[4]: *** [drivers/scsi/arm/arxescsi.o] Error 1
Signed-off-by: Arnd Bergmann <arnd at arndb.de>
diff --git a/drivers/scsi/arm/arxescsi.c b/drivers/scsi/arm/arxescsi.c
index a750aa7..689838a 100644
--- a/drivers/scsi/arm/arxescsi.c
+++ b/drivers/scsi/arm/arxescsi.c
@@ -80,21 +80,21 @@ static void arxescsi_pseudo_dma_write(unsigned char *addr, void __iomem *base)
" mov r1, %1\n"
" add r2, r1, #512\n"
" mov r4, #256\n"
- ".loop_1: ldmia r3!, {r6, r8, r10, r12}\n"
+ "1: ldmia r3!, {r6, r8, r10, r12}\n"
" mov r5, r6, lsl #16\n"
" mov r7, r8, lsl #16\n"
- ".loop_2: ldrb r0, [r1, #1536]\n"
+ "2: ldrb r0, [r1, #1536]\n"
" tst r0, #1\n"
- " beq .loop_2\n"
+ " beq 2b\n"
" stmia r2, {r5-r8}\n\t"
" mov r9, r10, lsl #16\n"
" mov r11, r12, lsl #16\n"
- ".loop_3: ldrb r0, [r1, #1536]\n"
+ "3: ldrb r0, [r1, #1536]\n"
" tst r0, #1\n"
- " beq .loop_3\n"
+ " beq 3b\n"
" stmia r2, {r9-r12}\n"
" subs r4, r4, #16\n"
- " bne .loop_1\n"
+ " bne 1b\n"
" ldmia sp!, {r0-r12}\n"
:
: "r" (addr), "r" (base));
More information about the linux-arm-kernel
mailing list