* zx279128s pinmux questions
@ 2026-11-20 19:24 Stefan Dösinger
2026-11-21 9:26 ` Agustin Santiago Isasmendi
0 siblings, 1 reply; 3+ messages in thread
From: Stefan Dösinger @ 2026-11-20 19:24 UTC (permalink / raw)
To: Navid Ghahremani, Agustin Santiago Isasmendi, linux-arm-kernel
[-- Attachment #1: Type: text/plain, Size: 959 bytes --]
Hi Navid, Agustin,
My next goal is to clean up my zx297520v3 pin control driver and write a
proper binding. Before I do that I would like to understand how pinctrl works
on your board to decide how much code reuse makes sense.
My main question is if the zx27 board can control individual pins or if
multiple pins need to be controlled together. On zx297520v3 I can set every
pin individually. Does the same apply to your boards?
The background: There are two ways to write pinctrl bindings and drivers:
Either exposing a simple pinmux=<int> setting, that sets pin 123 to a
specified function. The other way is to collect groups of pins and possible
functions they can be assigned to in the driver. They are controlled via
strings. I prefer to go via the former one, but as far as I understand it,
this is not possible if hardware links two or more pins together.
Can you advise me on what you know about your board's pin control?
Cheers,
Stefan
[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 870 bytes --]
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: zx279128s pinmux questions
2026-11-20 19:24 zx279128s pinmux questions Stefan Dösinger
@ 2026-11-21 9:26 ` Agustin Santiago Isasmendi
2026-11-21 10:16 ` Agustin Santiago Isasmendi
0 siblings, 1 reply; 3+ messages in thread
From: Agustin Santiago Isasmendi @ 2026-11-21 9:26 UTC (permalink / raw)
To: Stefan Dösinger; +Cc: Navid Ghahremani, linux-arm-kernel
Hi Stefan,
On the ZX279128S the pins are individually controllable.
The chip's manual has a table of the pads (GPIO 0-63, in order) with up
to four functions each; e.g. JTAG_TDO can be JTAG_TDO, SIM_VCCEN,
UART1_TXD or GPIO2. It doesn't give the register bits, so I mapped
those from the stock U-Boot (it writes a per-board table into the
block) and by flipping bits on my H3600 and watching the GPIO inputs:
- 0x94200000 + 0x00..0x0c hold the mux, one field per pin, 2 bits
where the pad has several functions and 1 bit (function or GPIO)
where it has one. E.g. bits 2k+1:2k of +0x00 are GPIO k for
k = 0..10, and +0x08 holds the NAND/SPI-NAND pads.
- +0x30..+0x38 have one bit per pin; clearing it makes the pin read 0,
so probably an input enable.
- +0x50..+0xa4 have a 6-bit config field per pin, five per word.
Bits 1:0 are the pull (01 up, 10 down); I haven't decoded the rest.
The layout is irregular, though (GPIO 11-21 sit in +0x04, GPIO 36-41 at
the top of +0x00, and the 1-bit pads in +0x0c are out of order), so a
driver needs a table per pin with the mux and config positions, much
like the old pinctrl-zx296718. With that table, a pinmux = <pin
function> binding like yours should work for us too.
I have the per-pin table (manual functions plus the register positions
found so far, some still unverified) if it's useful to you or Navid.
Agustin
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: zx279128s pinmux questions
2026-11-21 9:26 ` Agustin Santiago Isasmendi
@ 2026-11-21 10:16 ` Agustin Santiago Isasmendi
0 siblings, 0 replies; 3+ messages in thread
From: Agustin Santiago Isasmendi @ 2026-11-21 10:16 UTC (permalink / raw)
To: Stefan Dösinger; +Cc: Navid Ghahremani, linux-arm-kernel
Hi Stefan,
A follow-up to my earlier mail, with the details behind it. No mux
field on the ZX279128S controls more than one pin: each pin has its
own field, so nothing is linked together and a pinmux = <pin function>
binding works for us too. The fields are just not in pin order, so the
driver needs a table of where each pin's field is. This is what we
have so far.
Pin numbers are GPIO numbers; the functions come from the manual's pad
table, listed by mux value (0/1/2/3). For 1-bit fields, 0 = function
and 1 = GPIO. The last column says where each position comes from:
"hw" = flipped on my board and seen on the GPIO input, "code" = written
by the stock U-Boot or kernel for that pin, "inf" = inferred from the
manual and the board tables, not tested.
Mux, 0x94200000:
reg bits pins functions src
0x00 2k+1:2k GPIO0-10 e.g. GPIO2 = JTAG_TDO/SIM_VCCEN/ hw: 1-3,8,9
(k=0..10) UART1_TXD/GPIO2 code: 5
inf: rest
0x00 22, 23 GPIO36,37 UART0 RXD, TXD hw
0x00 24..27 GPIO38-41 I2C0 SCL/SDA, I2C1 SCL/SDA hw
0x04 1:0 GPIO11 SSP0_FS0 inf
0x04 3:2 GPIO12 SSP0_FS1 hw
0x04 5:4 GPIO14 SSP0_FS3 hw
0x04 7:6 GPIO15 SSP0_RXD code
0x04 9:8 GPIO16 SSP0_TXD hw
0x04 11:10.. GPIO17-21 SSP0_SCLK, PCM CLK/FS/TXD/RXD inf
19:18
0x04 21:20 GPIO13 SSP0_FS2 hw
0x04 23:22 GPIO42 RGMII TXCLK inf
0x08 2k+1:2k GPIO22-35 NAND ADQ0-7 and control pads; code
(k=0..13) 1 on the control pads = SPI-NAND
0x0c 0 GPIO53 PCIE0_nCLKREQ hw
0x0c 1 GPIO54 PCIE1_RSTB code
0x0c 2 GPIO55 PCIE1_nCLKREQ hw
0x0c 3..6 GPIO56-59 GEPHY LED0-3 hw
0x0c 7..11 GPIO43-47 RGMII TXCTL, TXD0-3 inf
0x0c 12, 13 GPIO50,51 TX/RX_PWRDOWN hw
0x0c 14 GPIO52 PCIE0_RSTB code
0x0c 15, 16 GPIO48,49 TX_SD, RX_LOS hw
0x0c 18 GPIO60 SDMMC_DATA0 hw
0x0c 19, 20 GPIO61,62 MDC0, MDIO0 code
0x0c 21 GPIO63 STRAP_NAND_PAGE_SIZE0 hw
Unknown: 0x00 bits 31:28, 0x04 bits 27:24, 0x08 bits 31:28, 0x0c
bit 17.
0x30..0x38, one bit per pin: bit n of 0x30 is GPIO n (hw for 1, 2, 3,
8, 9, 12-14, 28, 29); in 0x34 the order is mostly GPIO 32+n but shifts
by one from bit 18. Clearing a bit makes the pin read 0.
0x50..0xa4, 6-bit pad config, five fields per word. Three are tested
(0x54 field 0 = GPIO6, 0x60 fields 0 and 2 = GPIO18 and GPIO21). The
rest seem to follow the manual's pad order (JTAG at 0x50, UART0 at
0x74, I2C at 0x78, PCIe at 0x94/0x98, ...), which puts the pull-ups on
UART0_RXD, the I2C pads, the PCIe nCLKREQ pins and JTAG. Bits 1:0 are
the pull (01 up, 10 down); bits 5:2 are unknown.
Agustin
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2026-11-21 10:16 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2026-11-20 19:24 zx279128s pinmux questions Stefan Dösinger
2026-11-21 9:26 ` Agustin Santiago Isasmendi
2026-11-21 10:16 ` Agustin Santiago Isasmendi
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox