Am 05.02.24 um 02:45 schrieb Ping-Ke Shih:
-----Original Message----- From: Fiona Klute <fiona.klute@xxxxxx> Sent: Friday, February 2, 2024 8:11 PM To: linux-wireless@xxxxxxxxxxxxxxx; Ping-Ke Shih <pkshih@xxxxxxxxxxx> Cc: Kalle Valo <kvalo@xxxxxxxxxx>; Ulf Hansson <ulf.hansson@xxxxxxxxxx>; linux-mmc@xxxxxxxxxxxxxxx; Pavel Machek <pavel@xxxxxx>; Ondřej Jirman <megi@xxxxxx>; Fiona Klute <fiona.klute@xxxxxx> Subject: [PATCH 1/9] wifi: rtw88: Shared module for rtw8723x devices The already supported 8723d chip is very similar to 8703b/8723cs, split code that can be shared into a new module. The spec definition tables are combined into a struct so we only need one EXPORT_SYMBOL for them all. Signed-off-by: Fiona Klute <fiona.klute@xxxxxx> --- I'm not sure how who should be MODULE_AUTHOR in rtw8723x.c. Most of the code is from rtw8723d, I created the separate module. I don't want to claim authorship over code I didn't write, but assigning authorship unasked (by copying the MODULE_AUTHOR from rtw8723d.c) also seems wrong.If two MODULE_AUTHOR are allowed, maybe list both?
I just checked the macro documentation, it says to use multiple MODULE_AUTHOR for multiple authors, so I'll do that. :-)
[...]diff --git a/drivers/net/wireless/realtek/rtw88/rtw8723d.c b/drivers/net/wireless/realtek/rtw88/rtw8723d.c index c575476a00..68245f34b5 100644 --- a/drivers/net/wireless/realtek/rtw88/rtw8723d.c +++ b/drivers/net/wireless/realtek/rtw88/rtw8723d.c[...]@@ -930,6 +532,8 @@ static u8 rtw8723d_iqk_check_rx_failed(struct rtw_dev *rtwdev, return 0; } +#define IQK_LTE_WRITE_VAL 0x0000ff00 + static void rtw8723d_iqk_one_shot(struct rtw_dev *rtwdev, bool tx, const struct rtw_8723d_iqk_cfg *iqk_cfg) { @@ -937,7 +541,7 @@ static void rtw8723d_iqk_one_shot(struct rtw_dev *rtwdev, bool tx, /* enter IQK mode */ rtw_write32_mask(rtwdev, REG_FPGA0_IQK_11N, BIT_MASK_IQK_MOD, EN_IQK); - rtw8723d_iqk_config_lte_path_gnt(rtwdev); + rtw8723x_iqk_config_lte_path_gnt(rtwdev, IQK_LTE_WRITE_VAL);It would be better to discriminate IQK_LTE_WRITE_VAL_8723D and IQK_LTE_WRITE_VAL_8703B to prevent misreading.
Good point, I'll do that. Same for the ADDA_ON_VAL.
diff --git a/drivers/net/wireless/realtek/rtw88/rtw8723x.c b/drivers/net/wireless/realtek/rtw88/rtw8723x.c new file mode 100644 index 0000000000..2b58064547 --- /dev/null +++ b/drivers/net/wireless/realtek/rtw88/rtw8723x.c @@ -0,0 +1,561 @@[...]+const struct rtw8723x_common rtw8723x_common = { + .iqk_adda_regs = { + 0x85c, 0xe6c, 0xe70, 0xe74, 0xe78, 0xe7c, 0xe80, 0xe84, + 0xe88, 0xe8c, 0xed0, 0xed4, 0xed8, 0xedc, 0xee0, 0xeec + }, + .iqk_mac8_regs = {0x522, 0x550, 0x551}, + .iqk_mac32_regs = {0x40}, + .iqk_bb_regs = { + 0xc04, 0xc08, 0x874, 0xb68, 0xb6c, 0x870, 0x860, 0x864, 0xa04 + }, + + .ltecoex_addr = { + .ctrl = REG_LTECOEX_CTRL, + .wdata = REG_LTECOEX_WRITE_DATA, + .rdata = REG_LTECOEX_READ_DATA, + }, + .rf_sipi_addr = { + [RF_PATH_A] = { .hssi_1 = 0x820, .lssi_read = 0x8a0, + .hssi_2 = 0x824, .lssi_read_pi = 0x8b8}, + [RF_PATH_B] = { .hssi_1 = 0x828, .lssi_read = 0x8a4, + .hssi_2 = 0x82c, .lssi_read_pi = 0x8bc}, + }, + .dig = { + [0] = { .addr = 0xc50, .mask = 0x7f }, + [1] = { .addr = 0xc50, .mask = 0x7f }, + }, + .dig_cck = { + [0] = { .addr = 0xa0c, .mask = 0x3f00 }, + }, + .prioq_addrs = { + .prio[RTW_DMA_MAPPING_EXTRA] = { + .rsvd = REG_RQPN_NPQ + 2, .avail = REG_RQPN_NPQ + 3, + }, + .prio[RTW_DMA_MAPPING_LOW] = { + .rsvd = REG_RQPN + 1, .avail = REG_FIFOPAGE_CTRL_2 + 1, + }, + .prio[RTW_DMA_MAPPING_NORMAL] = { + .rsvd = REG_RQPN_NPQ, .avail = REG_RQPN_NPQ + 1, + }, + .prio[RTW_DMA_MAPPING_HIGH] = { + .rsvd = REG_RQPN, .avail = REG_FIFOPAGE_CTRL_2, + }, + .wsize = false, + }, + + .lck = __rtw8723x_lck, + .read_efuse = __rtw8723x_read_efuse, + .mac_init = __rtw8723x_mac_init, + .cfg_ldo25 = __rtw8723x_cfg_ldo25, + .set_tx_power_index = __rtw8723x_set_tx_power_index, + .efuse_grant = __rtw8723x_efuse_grant, + .false_alarm_statistics = __rtw8723x_false_alarm_statistics, + .iqk_backup_regs = __rtw8723x_iqk_backup_regs, + .iqk_restore_regs = __rtw8723x_iqk_restore_regs, + .iqk_similarity_cmp = __rtw8723x_iqk_similarity_cmp, + .pwrtrack_get_limit_ofdm = __rtw8723x_pwrtrack_get_limit_ofdm, + .pwrtrack_set_xtal = __rtw8723x_pwrtrack_set_xtal, + .coex_cfg_init = __rtw8723x_coex_cfg_init, + .fill_txdesc_checksum = __rtw8723x_fill_txdesc_checksum, +}; +EXPORT_SYMBOL(rtw8723x_common);I think these are just copy-and-paste stuff. Is there anything special you want me look deeper?
It should be just copy-and-paste. If you (or anyone else reading this) have the opportunity to test with an 8723D device to make sure I didn't make any mistake that breaks things there that'd be great.
+ +MODULE_AUTHOR("Fiona Klute <fiona.klute@xxxxxx>"); +MODULE_DESCRIPTION("Common functions for Realtek 802.11n wireless 8723x drivers"); +MODULE_LICENSE("Dual BSD/GPL"); diff --git a/drivers/net/wireless/realtek/rtw88/rtw8723x.h b/drivers/net/wireless/realtek/rtw88/rtw8723x.h new file mode 100644 index 0000000000..d3930f1f2c --- /dev/null +++ b/drivers/net/wireless/realtek/rtw88/rtw8723x.h[...]+/* IQK helper functions, defined as inline so they can be shared + * without needing an EXPORT_SYMBOL each. + */ +inline voidThese inline functions should be 'static'. (I believe you have addressed this, and will change them by v2.)
Yes, that was what broke building the driver into the kernel (not as a module) as reported by Pavel Machek.
+rtw8723x_iqk_backup_path_ctrl(struct rtw_dev *rtwdev, + struct rtw8723x_iqk_backup_regs *backup) +{ + backup->btg_sel = rtw_read8(rtwdev, REG_BTG_SEL); + rtw_dbg(rtwdev, RTW_DBG_RFK, "[IQK] original 0x67 = 0x%x\n", + backup->btg_sel); +} +[...]
