[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <55E55904.1070606@de.ibm.com>
Date: Tue, 01 Sep 2015 09:51:32 +0200
From: Christian Borntraeger <borntraeger@...ibm.com>
To: Alexander Kuleshov <kuleshovmail@...il.com>,
Martin Schwidefsky <schwidefsky@...ibm.com>,
Heiko Carstens <heiko.carstens@...ibm.com>
CC: linux390@...ibm.com, linux-s390@...r.kernel.org,
linux-kernel@...r.kernel.org
Subject: Re: [PATCH] [s390]: Use .rept directive to store/load fpu registers
Am 28.08.2015 um 15:49 schrieb Alexander Kuleshov:
> We can use .rept directive to simplify/decrease assembly code
> that makes storing and loading floating point registers in the
> arch/s390/kernel/swsusp.S.
>
> The resulting code after the patch is the same that before
>
> 2c4: 68 00 d2 00 ld %f0,512(%r13)
> 2c8: 68 10 d2 08 ld %f1,520(%r13)
> 2cc: 68 20 d2 10 ld %f2,528(%r13)
> 2d0: 68 30 d2 18 ld %f3,536(%r13)
> 2d4: 68 40 d2 20 ld %f4,544(%r13)
> 2d8: 68 50 d2 28 ld %f5,552(%r13)
> 2dc: 68 60 d2 30 ld %f6,560(%r13)
> 2e0: 68 70 d2 38 ld %f7,568(%r13)
> 2e4: 68 80 d2 40 ld %f8,576(%r13)
> 2e8: 68 90 d2 48 ld %f9,584(%r13)
> 2ec: 68 a0 d2 50 ld %f10,592(%r13)
> 2f0: 68 b0 d2 58 ld %f11,600(%r13)
> 2f4: 68 c0 d2 60 ld %f12,608(%r13)
> 2f8: 68 d0 d2 68 ld %f13,616(%r13)
> 2fc: 68 e0 d2 70 ld %f14,624(%r13)
> 300: 68 f0 d2 78 ld %f15,632(%r13)
>
> Signed-off-by: Alexander Kuleshov <kuleshovmail@...il.com>
> ---
> arch/s390/kernel/swsusp.S | 49 ++++++++++++++++-------------------------------
> 1 file changed, 17 insertions(+), 32 deletions(-)
>
> diff --git a/arch/s390/kernel/swsusp.S b/arch/s390/kernel/swsusp.S
> index ca62946..8011e14 100644
> --- a/arch/s390/kernel/swsusp.S
> +++ b/arch/s390/kernel/swsusp.S
> @@ -48,22 +48,15 @@ ENTRY(swsusp_arch_suspend)
> /* Store registers */
> mvc 0x318(4,%r1),__SF_EMPTY(%r15) /* move prefix to lowcore */
> stfpc 0x31c(%r1) /* store fpu control */
> - std 0,0x200(%r1) /* store f0 */
> - std 1,0x208(%r1) /* store f1 */
> - std 2,0x210(%r1) /* store f2 */
> - std 3,0x218(%r1) /* store f3 */
> - std 4,0x220(%r1) /* store f4 */
> - std 5,0x228(%r1) /* store f5 */
> - std 6,0x230(%r1) /* store f6 */
> - std 7,0x238(%r1) /* store f7 */
> - std 8,0x240(%r1) /* store f8 */
> - std 9,0x248(%r1) /* store f9 */
> - std 10,0x250(%r1) /* store f10 */
> - std 11,0x258(%r1) /* store f11 */
> - std 12,0x260(%r1) /* store f12 */
> - std 13,0x268(%r1) /* store f13 */
> - std 14,0x270(%r1) /* store f14 */
> - std 15,0x278(%r1) /* store f15 */
> +
> + /* store f0..f15 floating point registers */
> + .set i, 0
> + .set f, 0x200
> + .rept 16
> + std i, f(%r1)
> + .set i, (i+1)
> + .set f, (f+0x8)
> + .endr
I personally find the existing code easier to read, especially as it matches the
assembler output and I can directly where the slots are, (e.g. f11 is 0x258)
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists