[PATCH net-next] cxgb4: Revert "cxgb4: Remove SGE_HOST_PAGE_SIZE dependency on page size"

[PATCH net-next] cxgb4: Revert "cxgb4: Remove SGE_HOST_PAGE_SIZE dependency on page size"

[Vishal Kulkarni]

This reverts commit 2391b0030e241386d710df10e53e2cfc3c5d4fc1
SGE's BAR2 Doorbell/GTS Page Size is now interpreted correctly in the
firmware itself by using actual host page size. Hence previous commit
needs to be reverted.

Signed-off-by: Vishal Kulkarni <vishal@chelsio.com>
---
 drivers/net/ethernet/chelsio/cxgb4/t4_hw.c | 11 +++++++++++
 1 file changed, 11 insertions(+)

diff --git a/drivers/net/ethernet/chelsio/cxgb4/t4_hw.c b/drivers/net/ethernet/chelsio/cxgb4/t4_hw.c
index f9b70be..93feb25 100644
--- a/drivers/net/ethernet/chelsio/cxgb4/t4_hw.c
+++ b/drivers/net/ethernet/chelsio/cxgb4/t4_hw.c
@@ -7253,10 +7253,21 @@ int t4_fixup_host_params(struct adapter *adap, unsigned int page_size,
 			 unsigned int cache_line_size)
 {
 	unsigned int page_shift = fls(page_size) - 1;
+	unsigned int sge_hps = page_shift - 10;
 	unsigned int stat_len = cache_line_size > 64 ? 128 : 64;
 	unsigned int fl_align = cache_line_size < 32 ? 32 : cache_line_size;
 	unsigned int fl_align_log = fls(fl_align) - 1;
 
+	t4_write_reg(adap, SGE_HOST_PAGE_SIZE_A,
+		     HOSTPAGESIZEPF0_V(sge_hps) |
+		     HOSTPAGESIZEPF1_V(sge_hps) |
+		     HOSTPAGESIZEPF2_V(sge_hps) |
+		     HOSTPAGESIZEPF3_V(sge_hps) |
+		     HOSTPAGESIZEPF4_V(sge_hps) |
+		     HOSTPAGESIZEPF5_V(sge_hps) |
+		     HOSTPAGESIZEPF6_V(sge_hps) |
+		     HOSTPAGESIZEPF7_V(sge_hps));
+
 	if (is_t4(adap->params.chip)) {
 		t4_set_reg_field(adap, SGE_CONTROL_A,
 				 INGPADBOUNDARY_V(INGPADBOUNDARY_M) |
-- 
1.8.3.1

Re: [PATCH net-next] cxgb4: Revert "cxgb4: Remove SGE_HOST_PAGE_SIZE dependency on page size"

[David Miller]

From: Vishal Kulkarni <vishal@chelsio.com>
Date: Tue, 21 May 2019 09:12:02 +0530

> This reverts commit 2391b0030e241386d710df10e53e2cfc3c5d4fc1
> SGE's BAR2 Doorbell/GTS Page Size is now interpreted correctly in the
> firmware itself by using actual host page size. Hence previous commit
> needs to be reverted.
> 
> Signed-off-by: Vishal Kulkarni <vishal@chelsio.com>

Really confusing.

First of all, I see a lot of cxgb4 patch submissions targetting net-next that
are really legitimate bug fixes.

Are you only targetting net-next to be "on the safe side" because you are
unsure of the 'net' rules?  Please don't do that.  If it's a bug fix, send
it to 'net' as appropriate.

Second, so what happens to people running older firmware?  Have you made a
completely incompatible change to the firmware behavior?  If so, you have
to version check the firmware and use the correct interpretation based upon
how the firmware verion interprets things.

Thanks.

Re: [PATCH net-next] cxgb4: Revert "cxgb4: Remove SGE_HOST_PAGE_SIZE dependency on page size"

[Vishal Kulkarni]

Hi David,

On Wednesday, May 05/22/19, 2019 at 01:53:31 +0530, David Miller wrote:
> From: Vishal Kulkarni <vishal@chelsio.com>
> Date: Tue, 21 May 2019 09:12:02 +0530
> 
> > This reverts commit 2391b0030e241386d710df10e53e2cfc3c5d4fc1
> > SGE's BAR2 Doorbell/GTS Page Size is now interpreted correctly in the
> > firmware itself by using actual host page size. Hence previous commit
> > needs to be reverted.
> > 
> > Signed-off-by: Vishal Kulkarni <vishal@chelsio.com>
> 
> Really confusing.
> 
> First of all, I see a lot of cxgb4 patch submissions targetting net-next that
> are really legitimate bug fixes.
> 
> Are you only targetting net-next to be "on the safe side" because you are
> unsure of the 'net' rules?  Please don't do that.  If it's a bug fix, send
> it to 'net' as appropriate.

My apologies. I will retarget to 'net' and resend.

> 
> Second, so what happens to people running older firmware?  Have you made a
> completely incompatible change to the firmware behavior?  If so, you have
> to version check the firmware and use the correct interpretation based upon
> how the firmware verion interprets things.

The fix in the commit 2391b0030e241386d710df10e53e2cfc3c5d4fc1 is wrong
and introduced regression. The correct fix is done in the firmware and 
hence it needs to be reverted. The changes done in firmware is compatible.

> 
> Thanks.