2018-03-20 02:54:24

by Sinan Kaya

Subject: [PATCH v4 3/7] scsi: dpt_i2o: eliminate duplicate barriers on weakly-ordered archs

Code includes barrier() followed by writel(). writel() already has a
barrier
on some architectures like arm64.

This ends up CPU observing two barriers back to back before executing the
register write.

Since code already has an explicit barrier call, changing writel() to
writel_relaxed().

Signed-off-by: Sinan Kaya <[email protected]>
---
drivers/scsi/dpt_i2o.c | 6 +++---
1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/drivers/scsi/dpt_i2o.c b/drivers/scsi/dpt_i2o.c
index fd172b0..3c1e64a 100644
--- a/drivers/scsi/dpt_i2o.c
+++ b/drivers/scsi/dpt_i2o.c
@@ -1300,7 +1300,7 @@ static s32 adpt_i2o_post_this(adpt_hba* pHba, u32* data, int len)
wmb();

//post message
- writel(m, pHba->post_port);
+ writel_relaxed(m, pHba->post_port);
wmb();

return 0;
@@ -1390,7 +1390,7 @@ static s32 adpt_i2o_reset_hba(adpt_hba* pHba)

memcpy_toio(pHba->msg_addr_virt+m, msg, sizeof(msg));
wmb();
- writel(m, pHba->post_port);
+ writel_relaxed(m, pHba->post_port);
wmb();

while(*status == 0){
@@ -2797,7 +2797,7 @@ static s32 adpt_send_nop(adpt_hba*pHba,u32 m)
writel( 0,&msg[2]);
wmb();

- writel(m, pHba->post_port);
+ writel_relaxed(m, pHba->post_port);
wmb();
return 0;
}
--
2.7.4

2018-03-20 10:24:15

by Christoph Hellwig

Subject: Re: [PATCH v4 3/7] scsi: dpt_i2o: eliminate duplicate barriers on weakly-ordered archs

This is a basically unmaintained driver for historic hardware, and not the
right place for micro-optimizations.

2018-03-20 12:15:11

by Sinan Kaya

Subject: Re: [PATCH v4 3/7] scsi: dpt_i2o: eliminate duplicate barriers on weakly-ordered archs

On 2018-03-20 06:23, Christoph Hellwig wrote:
> This is a basically unmaintained driver for historic hardware, and not
> the
> right place for micro-optimizations.

Sure, I can drop this on the next version.