armv7-r/a: bugfix: fix a4 register use but not store in xxx_invalidate/flush/clean_all.S

[zhuyanlinzyl]

fix a4 register use but not store in xxx_invalidate/flush/clean_all.S

Signed-off-by: zhuyanlin zhuyanlin1@xiaomi.com

Summary

fix a4 register use but not store in xxx_invalidate/flush/clean_all.S

Impact

Testing

[pkarashchenko]

@zhuyanlinzyl do you have possibility to address comments this week?

[xiaoxiang781216]

We are in Spring Festival and come back to work in the next week.

[zhuyanlinzyl]

@masayuki2009 @pkarashchenko I update a new patch, please review again.

Use sub loop instead of add loop

And it work well in sabrelite qemu.

[zhuyanlinzyl]

hi, pkarashchenko , `And` is right, as it is `&=` `Add` is wrong, I fix it in new patch. You can review in this PR #5367

________________________________ From: Petro Karashchenko ***@***.***> Sent: 07 February 2022 16:50 To: apache/incubator-nuttx Cc: 朱琰琳; Mention Subject: [External Mail]Re: [apache/incubator-nuttx] armv7-r/a: bugfix: save and restore a4 register in xxx_dcache_all.S (PR #5367) *This message originated from outside of XIAOMI. Please treat this email with caution* @pkarashchenko commented on this pull request.

________________________________ In arch/arm/src/armv7-a/cp15_clean_dcache_all.S<#5367 (comment)>:

ldr r3, =0x7fff /* Isolate the NumSets field (bits 13-27) */

and r0, r3, r0, lsr #13 /* r0=NumSets (number of sets - 1) */ ldr r3, =0x3ff /* Isolate the way field (bits 3-12) */ - add r4, r3, r0, lsr #3 /* r4=(number of ways - 1) */ + add r4, r3, r4, lsr #3 /* r4=(number of ways - 1) */ I see that comment about and vs add is not answered. I would appreciate if author can write here the answer ― Reply to this email directly, view it on GitHub<#5367 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AU6JSNI4BU6KATB5KL5K6CLUZ6BTZANCNFSM5NAJOP4A>. Triage notifications on the go with GitHub Mobile for iOS<https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675> or Android<https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub>. You are receiving this because you were mentioned.Message ID: ***@***.***> #/******本邮件及其附件含有小米公司的保密信息，仅限于发送给上面地址中列出的个人或群组。禁止任何其他人以任何形式使用（包括但不限于全部或部分地泄露、复制、或散发）本邮件中的信息。如果您错收了本邮件，请您立即电话或邮件通知发件人并删除本邮件！ This e-mail and its attachments contain confidential information from XIAOMI, which is intended only for the person or entity whose address is listed above. Any use of the information contained herein in any way (including, but not limited to, total or partial disclosure, reproduction, or dissemination) by persons other than the intended recipient(s) is prohibited. If you receive this e-mail in error, please notify the sender by phone or email immediately and delete it!******/#

[masayuki2009]

@masayuki2009 @pkarashchenko I update a new patch, please review again.

Use sub loop instead of add loop
And it work well in sabrelite qemu.

@zhuyanlinzyl

Did you test with your board (I think Xiaomi uses dual Cortex-A7-based board) as well?

[zhuyanlinzyl]

@masayuki2009 Yes, I test in our smp a7 board, this patch can work

[masayuki2009]

@zhuyanlinzyl
Is this condition correct?
If the number of ways equals 4 (i.e. r1 = 3), it hits only 3times.

[zhuyanlinzyl]

@masayuki2009

Yes, sub and bne only loop 3times.
I use subs and bcs , It should loop 4times.

I update the patch. please review agian.

[masayuki2009]

@zhuyanlinzyl
subs?

[zhuyanlinzyl]

It's a typ mistake. I fix and update the patch @masayuki2009 .