Thread (12 messages) 12 messages, 2 authors, 2018-07-31

Re: [PATCH 3/9] ath9k: force rx_clear when disabling rx

From: Kalle Valo <hidden>
Date: 2018-07-30 19:43:40

Felix Fietkau [off-list ref] writes:
quoted hunk ↗ jump to hunk
From: Felix Fietkau <redacted>

This makes stopping Rx more reliable and should reduce the frequency of
Rx related DMA stop warnings. Don't use rx_clear in TX99 mode.

Signed-off-by: Felix Fietkau <nbd@nbd.name>
Signed-off-by: Helmut Schaa <redacted>
---
 drivers/net/wireless/ath/ath9k/mac.c | 9 +++++++--
 1 file changed, 7 insertions(+), 2 deletions(-)
diff --git a/drivers/net/wireless/ath/ath9k/mac.c b/drivers/net/wireless/ath/ath9k/mac.c
index 58d02c19b6d0..c9d2bf3fa135 100644
--- a/drivers/net/wireless/ath/ath9k/mac.c
+++ b/drivers/net/wireless/ath/ath9k/mac.c
@@ -678,13 +678,18 @@ void ath9k_hw_startpcureceive(struct ath_hw *ah, bool is_scanning)
 
 	ath9k_ani_reset(ah, is_scanning);
 
-	REG_CLR_BIT(ah, AR_DIAG_SW, (AR_DIAG_RX_DIS | AR_DIAG_RX_ABORT));
+	REG_CLR_BIT(ah, AR_DIAG_SW,
+		    AR_DIAG_RX_DIS | AR_DIAG_RX_ABORT | AR_DIAG_FORCE_RX_CLEAR);
 }
 EXPORT_SYMBOL(ath9k_hw_startpcureceive);
 
 void ath9k_hw_abortpcurecv(struct ath_hw *ah)
 {
-	REG_SET_BIT(ah, AR_DIAG_SW, AR_DIAG_RX_ABORT | AR_DIAG_RX_DIS);
+	u32 reg = AR_DIAG_RX_DIS | AR_DIAG_RX_ABORT;
+
+	if (!IS_ENABLED(CPTCFG_ATH9K_TX99))
+		reg |= AR_DIAG_FORCE_RX_CLEAR;
+	REG_SET_BIT(ah, AR_DIAG_SW, reg);
CPTCFG_ prefix is obviously wrong, so I'll drop this patch.

But I also wonder if IS_ENABLED() is really correct? Why would enabling
a Kconfig option change "normal mode" (= not tx99 mode) functionality,
even when nobody accesses the tx99 debugfs file? I would have thought
there's some kind of runtime check to see if tx99 mode is enabled or
something like that.

-- 
Kalle Valo
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help