[PATCH] arm64: Avoid breakage caused by .altmacro in fpsimd save/restore macros
Dave P Martin
Dave.Martin at arm.com
Thu Jan 29 08:51:00 PST 2015
On Thu, Jan 29, 2015 at 04:34:05PM +0000, Ard Biesheuvel wrote:
> On 29 January 2015 at 16:24, Dave Martin <Dave.Martin at arm.com> wrote:
> > Alternate macro mode is not a property of a macro definition, but a
> > gas runtime state that alters the way macros are expanded for ever
> > after (until .noaltmacro is seen).
> >
> > This means that subsequent assembly code that calls other macros can
> > break if fpsimdmacros.h is included.
> >
> > Since these instruction sequences are simple (if dull -- but in a
> > good way), this patch solves the problem by simply expanding the
> > .irp loops. The pre-existing fpsimd_{save,restore} macros weren't
> > rolled with .irp anyway and the sequences affected are short, so
> > this change restores consistency at little cost.
> >
> > Signed-off-by: Dave Martin <Dave.Martin at arm.com>
>
> Acked-by: Ard Biesheuvel <ard.biesheuvel at linaro.org>
Thanks
(Also, apologies for accidentally missing you from cc)
Cheers
---Dave
>
> > ---
> > arch/arm64/include/asm/fpsimdmacros.h | 43 ++++++++++++++++++++++++---------
> > 1 file changed, 32 insertions(+), 11 deletions(-)
> >
> > diff --git a/arch/arm64/include/asm/fpsimdmacros.h b/arch/arm64/include/asm/fpsimdmacros.h
> > index 007618b..a2daf12 100644
> > --- a/arch/arm64/include/asm/fpsimdmacros.h
> > +++ b/arch/arm64/include/asm/fpsimdmacros.h
> > @@ -76,7 +76,6 @@
> > fpsimd_restore_fpcr x\tmpnr, \state
> > .endm
> >
> > -.altmacro
> > .macro fpsimd_save_partial state, numnr, tmpnr1, tmpnr2
> > mrs x\tmpnr1, fpsr
> > str w\numnr, [\state, #8]
> > @@ -86,11 +85,22 @@
> > add \state, \state, x\numnr, lsl #4
> > sub x\tmpnr1, x\tmpnr1, x\numnr, lsl #1
> > br x\tmpnr1
> > - .irp qa, 30, 28, 26, 24, 22, 20, 18, 16, 14, 12, 10, 8, 6, 4, 2, 0
> > - .irp qb, %(qa + 1)
> > - stp q\qa, q\qb, [\state, # -16 * \qa - 16]
> > - .endr
> > - .endr
> > + stp q30, q31, [\state, #-16 * 30 - 16]
> > + stp q28, q29, [\state, #-16 * 28 - 16]
> > + stp q26, q27, [\state, #-16 * 26 - 16]
> > + stp q24, q25, [\state, #-16 * 24 - 16]
> > + stp q22, q23, [\state, #-16 * 22 - 16]
> > + stp q20, q21, [\state, #-16 * 20 - 16]
> > + stp q18, q19, [\state, #-16 * 18 - 16]
> > + stp q16, q17, [\state, #-16 * 16 - 16]
> > + stp q14, q15, [\state, #-16 * 14 - 16]
> > + stp q12, q13, [\state, #-16 * 12 - 16]
> > + stp q10, q11, [\state, #-16 * 10 - 16]
> > + stp q8, q9, [\state, #-16 * 8 - 16]
> > + stp q6, q7, [\state, #-16 * 6 - 16]
> > + stp q4, q5, [\state, #-16 * 4 - 16]
> > + stp q2, q3, [\state, #-16 * 2 - 16]
> > + stp q0, q1, [\state, #-16 * 0 - 16]
> > 0:
> > .endm
> >
> > @@ -103,10 +113,21 @@
> > add \state, \state, x\tmpnr2, lsl #4
> > sub x\tmpnr1, x\tmpnr1, x\tmpnr2, lsl #1
> > br x\tmpnr1
> > - .irp qa, 30, 28, 26, 24, 22, 20, 18, 16, 14, 12, 10, 8, 6, 4, 2, 0
> > - .irp qb, %(qa + 1)
> > - ldp q\qa, q\qb, [\state, # -16 * \qa - 16]
> > - .endr
> > - .endr
> > + ldp q30, q31, [\state, #-16 * 30 - 16]
> > + ldp q28, q29, [\state, #-16 * 28 - 16]
> > + ldp q26, q27, [\state, #-16 * 26 - 16]
> > + ldp q24, q25, [\state, #-16 * 24 - 16]
> > + ldp q22, q23, [\state, #-16 * 22 - 16]
> > + ldp q20, q21, [\state, #-16 * 20 - 16]
> > + ldp q18, q19, [\state, #-16 * 18 - 16]
> > + ldp q16, q17, [\state, #-16 * 16 - 16]
> > + ldp q14, q15, [\state, #-16 * 14 - 16]
> > + ldp q12, q13, [\state, #-16 * 12 - 16]
> > + ldp q10, q11, [\state, #-16 * 10 - 16]
> > + ldp q8, q9, [\state, #-16 * 8 - 16]
> > + ldp q6, q7, [\state, #-16 * 6 - 16]
> > + ldp q4, q5, [\state, #-16 * 4 - 16]
> > + ldp q2, q3, [\state, #-16 * 2 - 16]
> > + ldp q0, q1, [\state, #-16 * 0 - 16]
> > 0:
> > .endm
> > --
> > 1.7.10.4
> >
> >
> >
> > _______________________________________________
> > linux-arm-kernel mailing list
> > linux-arm-kernel at lists.infradead.org
> > http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
>
More information about the linux-arm-kernel
mailing list