Skip to contentSkip to conversation

staging: rtl8723bs: replace magic numbers with named constants

PATCH In review Adith-Joshua 3 participants 6 messages Apr 11 – Apr 13 2 messages not in the archive Staging

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 staging: rtl8723bs: replace magic numbers with named constants

Replace hardcoded magic numbers in rtl8723b_InitBeaconParameters() with descriptive macros to improve readability and maintainability.

No functional changes.

---
 .../staging/rtl8723bs/hal/rtl8723b_hal_init.c   | 17 ++++++++++++++---
 1 file changed, 14 insertions(+), 3 deletions(-)
drivers/staging/rtl8723bs/hal/rtl8723b_hal_init.c +14 -3
index 8d259820f103..41f561120af5 100644
@@ -864,6 +864,18 @@ void rtl8723b_read_chip_version(struct adapter *padapter)
 	ReadChipVersion8723B(padapter);
 }
 
+#define TBTT_PROHIBIT_HOLD_TIME_MS   0x64
+#define TBTT_PROHIBIT_SETUP_TIME_MS  0x04
+
+#define TBTT_PROHIBIT_VALUE \
+	((TBTT_PROHIBIT_HOLD_TIME_MS << 8) | TBTT_PROHIBIT_SETUP_TIME_MS)
+
+#define BCNTCFG_AIFS 0x66
+#define BCNTCFG_CW   0x0F
+
+#define BCNTCFG_VALUE \
+	((BCNTCFG_AIFS << 8) | BCNTCFG_CW)
+
 void rtl8723b_InitBeaconParameters(struct adapter *padapter)
 {
 	struct hal_com_data *pHalData = GET_HAL_DATA(padapter);
@@ -878,8 +890,7 @@ void rtl8723b_InitBeaconParameters(struct adapter *padapter)
 
 	rtw_write16(padapter, REG_BCN_CTRL, val16);
 
-	/*  TODO: Remove these magic number */
-	rtw_write16(padapter, REG_TBTT_PROHIBIT, 0x6404);/*  ms */
+	rtw_write16(padapter, REG_TBTT_PROHIBIT, TBTT_PROHIBIT_VALUE);/*  ms */
 	/*  Firmware will control REG_DRVERLYINT when power saving is enable, */
 	/*  so don't set this register on STA mode. */
 	if (check_fwstate(&padapter->mlmepriv, WIFI_STATION_STATE) == false)
@@ -888,7 +899,7 @@ void rtl8723b_InitBeaconParameters(struct adapter *padapter)
 
 	/*  Suggested by designer timchen. Change beacon AIFS to the largest number */
 	/*  because test chip does not contension before sending beacon. by tynli. 2009.11.03 */
-	rtw_write16(padapter, REG_BCNTCFG, 0x660F);
+	rtw_write16(padapter, REG_BCNTCFG, BCNTCFG_VALUE);
 
 	pHalData->RegBcnCtrlVal = rtw_read8(padapter, REG_BCN_CTRL);
 	pHalData->RegTxPause = rtw_read8(padapter, REG_TXPAUSE);
Signature · 1 line
2.53.0
#1 lore raw
Quoted 7 lines · Adith-Joshua

Replace hardcoded magic numbers in rtl8723b_InitBeaconParameters() with descriptive macros to improve readability and maintainability.

No functional changes.

---

Where did you find these numbers? Presumably there is a spec somewhere? What does BCNTCFG_CW mean? This really looks like an AI patch...

regards,
dan carpenter

#2 lore raw
Quoted 9 lines · Dan Carpenter
Quoted 7 lines · Adith-Joshua

Replace hardcoded magic numbers in rtl8723b_InitBeaconParameters() with descriptive macros to improve readability and maintainability.

No functional changes.

---

Quoted 6 lines

Where did you find these numbers? Presumably there is a spec somewhere? What does BCNTCFG_CW mean? This really looks like an AI patch...

regards,
dan carpenter

Hi Dan,
I believe you are right about this patch being written by AI. As I noted in my other reply, if you cross-reference these registers with the primary upstream driver (drivers/net/wireless/realtek/rtw88/reg.h), the field mappings introduced in this patch are actually backwards. It looks like an AI saw the 0x660F magic number and hallucinated that 0x66 was the aifs and 0x0F was the cw, when in reality aifs occupies the lower 8 bits. It also hallucinates an _MS suffix based on a loose code comment, completely ignoring that 802.11 beacon offsets use time units (TU). Best regards,
Luka Gejak

A message that is not in the archive (sent off-list or not yet received): CAFvcH2CjS=HMxt_Dd04eMAELQtxdB5LYmLa_1TLnJeEuKvR6=Q@mail.gmail.com

#4 lore raw
Quoted 10 lines · Adith Joshua

Hi Dan,

Thanks for taking a look at my patch.

These values come from the RTL8723BS register layout(as in original vendor driver)

For TBTT_PROHIBIT (0x6404), the register is split into two fields. The upper byte(0x64) is the TBTT prohibit hold time. The lower byte(0x04) is the setup/margin time before TBTT

Is this available online somewhere? I did manage to find TBTT_PROHIBIT on google for the rtl8188e driver and the explanation is here:

https://android.googlesource.com/kernel/msm/+/android-7.1.0_r0.2/drivers/staging/rtl8188eu/include/rtl8188e_spec.h#296

It says that the "TBTT prohibit setup in unit of 32us." but the patch says it is in units of ms so that's wrong.

Quoted 4 lines

For BCNTCFG(0x660F) the upper byte(0x66) configures AIFS value for beacon transmission. The lower byte(0x0F) sets the contention window to the max value

I wasn't able to verify any of this. Please provide us a link to your original vendor driver.

Quoted 7 lines

The intention of this change was to replace the unclear hardcoded values with proper macros(like the todo said)

If it is needed I’ll send a v2 with more explicit names and brief comments for the bit fields

*grumble* It's honestly, still not clear if you used AI to write the patch or not.

regards,
dan carpenter

A message that is not in the archive (sent off-list or not yet received): CAFvcH2DQJtB6B9L+SMgg_aPXGe96hw2i6+C1N0uhapHvJQHd2w@mail.gmail.com

#6 lore raw
Quoted 14 lines · Adith Joshua

Hi Dan, Luka,

Thanks for the review.

Made some careless errors there... sorry about that

Looking at it again, I see that the BCNTCFG field mapping in my patch is wrong (I swapped AIFS and CW in the naming). The TBTT_PROHIBIT suffix was also an assumption based on a vague comment, which I shouldn’t have relied on.

I was looking at:
https://android.googlesource.com/kernel/msm/+/android-7.1.0_r0.2/drivers/staging/rtl8188eu/include/rtl8188e_spec.h#296

Quoted 8 lines

and also cross-referencing with rtw88:
drivers/net/wireless/realtek/rtw88/reg.h

And no, this wasn’t AI generated 🙂(I know the choice of emoji could be better)

I'll send a v2 with these corrections.

Please don't bother.

This explanation is again far too vague to be checked. I would prefer to leave it as-is until we are more sure that the information is correct. Misleading documentation is worse than having none.

regards,
dan carpenter

#7 lore raw
In reply to Adith-Joshua
Quoted 15 lines · Adith-Joshua

Replace hardcoded magic numbers in rtl8723b_InitBeaconParameters() with descriptive macros to improve readability and maintainability.

No functional changes.

---
 .../staging/rtl8723bs/hal/rtl8723b_hal_init.c   | 17 ++++++++++++++---
 1 file changed, 14 insertions(+), 3 deletions(-)
drivers/staging/rtl8723bs/hal/rtl8723b_hal_init.c +0 -0
index 8d259820f103..41f561120af5 100644

drivers/staging/rtl8723bs/hal/rtl8723b_hal_init.c void rtl8723b_read_chip_version(struct adapter *padapter) · line 868

Quoted 5 lines
drivers/staging/rtl8723bs/hal/rtl8723b_hal_init.c +2 -0
@@ -864,6 +864,18 @@ void rtl8723b_read_chip_version(struct adapter *padapter)
 	ReadChipVersion8723B(padapter);
 }
 
+#define TBTT_PROHIBIT_HOLD_TIME_MS   0x64
+#define TBTT_PROHIBIT_SETUP_TIME_MS  0x04

Please drop the _MS suffix here. While the legacy comment says "ms", 802.11 beacon offsets are almost universally measured in time units (TU, which are 1024 microseconds). 0x64 is 100 TU. Baking "MS" into the macro name based on a loose comment introduces false precision. Please fix the field mappings and send a v2. I highly recommend taking a look at how rtw88 handles these exact same registers
(REG_BCNTCFG and REG_TBTT_PROHIBIT) and trying to align your macros with their structure where possible.

drivers/staging/rtl8723bs/hal/rtl8723b_hal_init.c void rtl8723b_read_chip_version(struct adapter *padapter) · line 877

Quoted 9 lines
drivers/staging/rtl8723bs/hal/rtl8723b_hal_init.c +9 -0
+
+#define TBTT_PROHIBIT_VALUE \
+	((TBTT_PROHIBIT_HOLD_TIME_MS << 8) | TBTT_PROHIBIT_SETUP_TIME_MS)
+
+#define BCNTCFG_AIFS 0x66
+#define BCNTCFG_CW   0x0F
+
+#define BCNTCFG_VALUE \
+	((BCNTCFG_AIFS << 8) | BCNTCFG_CW)

This is completely backwards. If you cross-reference this with the primary upstream rtw88 driver (drivers/net/wireless/realtek/rtw88/reg.h) , aifs is actually the lower 8 bits (GENMASK(7, 0)) and cw is in the upper bits. So 0x0F is the aifs (a typical max value of 15), and 0x66 represents the cwmax and cwmin values. While your macro happens to evaluate to the original 0x660F magic number, the field names are swapped and would actively mislead future developers.

Quoted 15 lines
drivers/staging/rtl8723bs/hal/rtl8723b_hal_init.c +2 -2
+
 void rtl8723b_InitBeaconParameters(struct adapter *padapter)
 {
 	struct hal_com_data *pHalData = GET_HAL_DATA(padapter);
@@ -878,8 +890,7 @@ void rtl8723b_InitBeaconParameters(struct adapter *padapter)
 
 	rtw_write16(padapter, REG_BCN_CTRL, val16);
 
-	/*  TODO: Remove these magic number */
-	rtw_write16(padapter, REG_TBTT_PROHIBIT, 0x6404);/*  ms */
+	rtw_write16(padapter, REG_TBTT_PROHIBIT, TBTT_PROHIBIT_VALUE);/*  ms */
 	/*  Firmware will control REG_DRVERLYINT when power saving is enable, */
 	/*  so don't set this register on STA mode. */
 	if (check_fwstate(&padapter->mlmepriv, WIFI_STATION_STATE) == false)

drivers/staging/rtl8723bs/hal/rtl8723b_hal_init.c void rtl8723b_InitBeaconParameters(struct adapter *padapter) · line 905

Quoted 8 lines
drivers/staging/rtl8723bs/hal/rtl8723b_hal_init.c +1 -1
@@ -888,7 +899,7 @@ void rtl8723b_InitBeaconParameters(struct adapter *padapter)
 
 	/*  Suggested by designer timchen. Change beacon AIFS to the largest number */
 	/*  because test chip does not contension before sending beacon. by tynli. 2009.11.03 */
-	rtw_write16(padapter, REG_BCNTCFG, 0x660F);
+	rtw_write16(padapter, REG_BCNTCFG, BCNTCFG_VALUE);
 
 	pHalData->RegBcnCtrlVal = rtw_read8(padapter, REG_BCN_CTRL);
 	pHalData->RegTxPause = rtw_read8(padapter, REG_TXPAUSE);

Best regards,
Luka Gejak