Skip to contentSkip to conversation

Commodore's Gayle pcmcia driver

RFC PATCH In review Paolo Pisati 2 participants 6 messages Jan 12, 2025 – Jan 16, 2025

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

cover Commodore's Gayle pcmcia driver

Here is a second attempt to get the PCMCIA side of Gayle's to work: it merges my previous attempt[1], some historical (e.g. v2.6.10 era) work from Kars de Jong[2](the set_io_map() and set_mem_map() are a blatant cut&paste), and it follows the NetBSD code[3] (when it made sense) - credit goes where credit is due.

The good news first:

if you have a pcnet_cs or fmvj18x_cs PCMCIA eth card, chances are it will work (i've 3 of them working here).

And now the bad one:

1) 3c589_cs (3COM Etherlink III): card is detected, it receives incoming packets (eth0 counter increases) but TX fails

2) axnet_cs: card is detected, but as soon as i try to up it, it errors out ("axnet_cs 0.0 eth0: Hw. address read/write mismap X")

3) pata_pcmcia (no name CF to PCMCIA adapter): it's recognized, but fails to attach ("pata_pcmcia 0.0: probe with driver pata_pcmcia failed with error -12") - looking a bit closer, it fails in pcmcia_init_one() when:

io_addr = devm_ioport_map(&pdev->dev, io_base, 8);

just notice that devm_ioport_unmap() is not implemented if CONFIG_HAS_IOPORT_MAP is not set (and M68K doesn't implement it).

And, finally, the extremly sad part:

1) i've forced the word and long I/O helpers to be little endian, and this is clearly a big no (see io_mm.h).

2) for the love of me, i couldn't figure out the proper way of acking an incoming interrupt on IRQ_AMIGA_PORTS at Gayle's level (and subsequenlty call all PCMCIA's card handlers registered), so i resorted to an ugly hack: i registered a dummy hanlder on IRQ_AMIGA_PORTS, checked for GAYLE_IRQ_IRQ, cleared the interrupt at Gayle's level, and then returned IRQ_NONE so the remaining handlers registered on IRQ_AMIGA_PORTS could be invoked - see gayle_irq_dummy().

3) the get_status() and set_socket() are pretty barebone, but that is intentional since, so far, i couldn't find a case where implementing all SS states made any difference.

If you want to see what's going on under the hood, stick a:

ccflags-y += -DDEBUG

in drivers/pcmcia/Makefile and rebuild your kernel.

Let me know what you think.

1: https://www.spinics.net/lists/linux-m68k/msg25982.html
2: https://dpaste.com//GYSLEKLLM and https://dpaste.com//EQE56PPSU
3: https://github.com/NetBSD/src/blob/trunk/sys/arch/amiga/dev/gayle_pcmcia.c

Paolo Pisati (1):
pcmcia: socket driver for Commodore's Gayle

 arch/m68k/amiga/pcmcia.c          |  59 ++++++
 arch/m68k/amiga/platform.c        |  29 +++
 arch/m68k/include/asm/amipcmcia.h |  20 ++
 arch/m68k/include/asm/io_mm.h     |  20 +-
 drivers/pcmcia/Kconfig            |   9 +
 drivers/pcmcia/Makefile           |   1 +
 drivers/pcmcia/gayle.c            | 303 ++++++++++++++++++++++++++++++
 7 files changed, 437 insertions(+), 4 deletions(-)
 create mode 100644 drivers/pcmcia/gayle.c
Signature · 1 line
2.34.1
#1 lore raw

PATCH 1/1 pcmcia: socket driver for Commodore's Gayle

---
 arch/m68k/amiga/pcmcia.c          |  59 ++++++
 arch/m68k/amiga/platform.c        |  29 +++
 arch/m68k/include/asm/amipcmcia.h |  20 ++
 arch/m68k/include/asm/io_mm.h     |  20 +-
 drivers/pcmcia/Kconfig            |   9 +
 drivers/pcmcia/Makefile           |   1 +
 drivers/pcmcia/gayle.c            | 306 ++++++++++++++++++++++++++++++
 7 files changed, 440 insertions(+), 4 deletions(-)
 create mode 100644 drivers/pcmcia/gayle.c
arch/m68k/amiga/pcmcia.c +59 -0
index 63cce6b590df..21dcd2719319 100644
@@ -13,20 +13,37 @@
 */
 
 #include <linux/types.h>
+#include <linux/delay.h>
 #include <linux/jiffies.h>
 #include <linux/timer.h>
 #include <linux/module.h>
 
+#include <pcmcia/ss.h>
 #include <asm/amigayle.h>
 #include <asm/amipcmcia.h>
 
 /* gayle config byte for program voltage and access speed */
 static unsigned char cfg_byte = GAYLE_CFG_0V|GAYLE_CFG_150NS;
 
+/* PCMCIA I/O maps */
+static struct pccard_io_map gayle_io_maps[MAX_IO_WIN];
+
+/*
+ * according to NetBSD (commit e26e7a8a2278 and 100db321d09e in [1])
+ * and depending on Gayle's revision, there are two methods for a PCMCIA
+ * soft-reset
+ *
+ * 1: https://github.com/NetBSD/src.git
+ */
 void pcmcia_reset(void)
 {
 	unsigned long reset_start_time = jiffies;
 
+	gayle.intreq = 0xff;
+	udelay(10);
+	gayle.intreq = 0xfc;
+	udelay(20);
+
 	gayle_reset = 0x00;
 	while (time_before(jiffies, reset_start_time + 1*HZ/100));
 	READ_ONCE(gayle_reset);
@@ -65,6 +82,11 @@ int pcmcia_copy_tuple(unsigned char tuple_id, void *tuple, int max_len)
 }
 EXPORT_SYMBOL(pcmcia_copy_tuple);
 
+unsigned char pcmcia_get_voltage(void) {
+	return gayle.config & 0x03;
+}
+EXPORT_SYMBOL(pcmcia_get_voltage);
+
 void pcmcia_program_voltage(int voltage)
 {
 	unsigned char v;
@@ -119,3 +141,40 @@ void pcmcia_write_disable(void)
 }
 EXPORT_SYMBOL(pcmcia_write_disable);
 
+void gayle_set_io_win(int win, unsigned char flags, unsigned int start, unsigned int stop)
+{
+	struct pccard_io_map *map = &gayle_io_maps[win];
+
+	map->flags = flags;
+	map->start = start;
+	map->stop = stop;
+}
+
+unsigned long gayle_get_byte_base(unsigned long port)
+{
+	struct pccard_io_map *map;
+	int i;
+
+	/* Simple case first */
+	if (!(port & 1)) {
+		return (GAYLE_IO + port);
+	}
+
+	for (i = 0, map = gayle_io_maps; i < MAX_IO_WIN; i++, map++) {
+		if ((map->flags & MAP_ACTIVE) &&
+		    (port >= map->start) && (port <= map->stop)) {
+			if (map->flags & MAP_16BIT) {
+				return (GAYLE_IO + port);
+			} else {
+				/* Assume MAP_AUTOSZ works this way */
+				return (GAYLE_IO + port + ((port & 1) * GAYLE_ODD));
+			}
+		}
+	}
+
+	/* Sometimes an I/O access is done after a card has been removed. */
+	return (GAYLE_IO + port);
+}
+
+EXPORT_SYMBOL(gayle_set_io_win);
+EXPORT_SYMBOL(gayle_get_byte_base);
arch/m68k/amiga/platform.c +29 -0
index d34029d7b058..aff2dcfbcc98 100644
@@ -110,6 +110,26 @@ static const struct gayle_ide_platform_data a1200_ide_pdata __initconst = {
 	.explicit_ack	= 1,
 };
 
+static const struct resource a1200_pcmcia_resource[] __initconst = {
+	[0] = {
+		.name	= "Gayle memory",
+		.start	= GAYLE_RAM,
+		.end	= GAYLE_RAM+GAYLE_RAMSIZE-1,
+		.flags	= IORESOURCE_MEM,
+	},
+	[1] = {
+		.name	= "Gayle attribute",
+		.start	= GAYLE_ATTRIBUTE,
+		.end	= GAYLE_ATTRIBUTE+GAYLE_ATTRIBUTESIZE-1,
+		.flags	= IORESOURCE_MEM,
+	},
+	[2] = {
+		.name	= "Gayle I/O",
+		.start	= GAYLE_IO,
+		.end	= GAYLE_IO+GAYLE_IOSIZE-1,
+		.flags	= IORESOURCE_MEM,
+	}
+};
 
 static const struct resource a4000_ide_resource __initconst = {
 	.start	= 0xdd2000,
@@ -190,6 +210,15 @@ static int __init amiga_init_devices(void)
 						 sizeof(a1200_ide_pdata));
 		if (error)
 			return error;
+
+	}
+
+	if (AMIGAHW_PRESENT(PCMCIA)) {
+		pdev = platform_device_register_simple("amiga-gayle-pcmcia", -1,
+						       a1200_pcmcia_resource,
+						       ARRAY_SIZE(a1200_pcmcia_resource));
+		if (IS_ERR(pdev))
+			return PTR_ERR(pdev);
 	}
 
 	if (AMIGAHW_PRESENT(A4000_IDE)) {
arch/m68k/include/asm/amipcmcia.h +20 -0
index 6f1ec1887d82..7df336934fb2 100644
@@ -19,10 +19,14 @@
 
 void pcmcia_reset(void);
 int pcmcia_copy_tuple(unsigned char tuple_id, void *tuple, int max_len);
+unsigned char pcmcia_get_voltage(void);
 void pcmcia_program_voltage(int voltage);
 void pcmcia_access_speed(int speed);
 void pcmcia_write_enable(void);
 void pcmcia_write_disable(void);
+void gayle_set_io_win(int win, unsigned char flags, unsigned int start, unsigned
+		      int stop);
+unsigned long gayle_get_byte_base(unsigned long port);
 
 static inline u_char pcmcia_read_status(void)
 {
@@ -39,6 +43,12 @@ static inline void pcmcia_ack_int(u_char intreq)
 	gayle.intreq = 0xf8;
 }
 
+static inline void pcmcia_ack_ccdet(void)
+{
+	/* reset GAYLE_IRQ_CCDET */
+	gayle.intreq = 0xbb;
+}
+
 static inline void pcmcia_enable_irq(void)
 {
 	gayle.inten |= GAYLE_IRQ_IRQ;
@@ -51,6 +61,16 @@ static inline void pcmcia_disable_irq(void)
 
 #define PCMCIA_INSERTED	(gayle.cardstatus & GAYLE_CS_CCDET)
 
+static inline void pcmcia_disable_ccdet_irq(void)
+{
+	gayle.inten &= ~GAYLE_IRQ_CCDET;
+}
+
+static inline u_char pcmcia_wena(void)
+{
+	return (gayle.cardstatus & GAYLE_CS_WR);
+}
+
 /* valid voltages for pcmcia_ProgramVoltage */
 
 #define PCMCIA_0V	0
arch/m68k/include/asm/io_mm.h +16 -4
index 090aec54b8fa..567048692999 100644
@@ -56,7 +56,9 @@
 #ifdef CONFIG_AMIGA_PCMCIA
 #include <asm/amigayle.h>
 
-#define AG_ISA_IO_B(ioaddr) ( GAYLE_IO+(ioaddr)+(((ioaddr)&1)*GAYLE_ODD) )
+extern unsigned long gayle_get_byte_base(unsigned long);
+
+#define AG_ISA_IO_B(ioaddr) ( gayle_get_byte_base(ioaddr) )
 #define AG_ISA_IO_W(ioaddr) ( GAYLE_IO+(ioaddr) )
 
 #ifndef MULTI_ISA
@@ -199,20 +201,30 @@ static inline u16 __iomem *isa_mtw(unsigned long addr)
 
 
 #define isa_inb(port)      in_8(isa_itb(port))
+#define isa_outb(val,port) out_8(isa_itb(port),(val))
+#define isa_readb(p)       in_8(isa_mtb((unsigned long)(p)))
+#define isa_writeb(val,p)  out_8(isa_mtb((unsigned long)(p)),(val))
+
+#ifdef CONFIG_PCMCIA_GAYLE
+#define isa_inw(port)	   in_le16(isa_itw(port))
+#define isa_inl(port)	   in_le32(isa_itl(port))
+#define isa_outw(val,port) out_le16(isa_itw(port),(val))
+#define isa_outl(val,port) out_le32(isa_itl(port),(val))
+#define isa_readw(p)	   in_le16(isa_mtw((unsigned long)(p)))
+#define isa_writew(val,p)  out_le16(isa_mtw((unsigned long)(p)),(val))
+#else
 #define isa_inw(port)      (ISA_SEX ? in_be16(isa_itw(port)) : in_le16(isa_itw(port)))
 #define isa_inl(port)      (ISA_SEX ? in_be32(isa_itl(port)) : in_le32(isa_itl(port)))
-#define isa_outb(val,port) out_8(isa_itb(port),(val))
 #define isa_outw(val,port) (ISA_SEX ? out_be16(isa_itw(port),(val)) : out_le16(isa_itw(port),(val)))
 #define isa_outl(val,port) (ISA_SEX ? out_be32(isa_itl(port),(val)) : out_le32(isa_itl(port),(val)))
 
-#define isa_readb(p)       in_8(isa_mtb((unsigned long)(p)))
 #define isa_readw(p)       \
 	(ISA_SEX ? in_be16(isa_mtw((unsigned long)(p)))	\
 		 : in_le16(isa_mtw((unsigned long)(p))))
-#define isa_writeb(val,p)  out_8(isa_mtb((unsigned long)(p)),(val))
 #define isa_writew(val,p)  \
 	(ISA_SEX ? out_be16(isa_mtw((unsigned long)(p)),(val))	\
 		 : out_le16(isa_mtw((unsigned long)(p)),(val)))
+#endif
 
 #ifdef CONFIG_ATARI_ROM_ISA
 #define isa_rom_inb(port)      rom_in_8(isa_itb(port))
drivers/pcmcia/Kconfig +9 -0
index dddb235dd020..62da9717d78e 100644
@@ -253,4 +253,13 @@ config PCCARD_NONSTATIC
 config PCCARD_IODYN
 	bool
 
+config PCMCIA_GAYLE
+	tristate "Amiga Gayle bridge support"
+	depends on PCMCIA && AMIGA && AMIGA_PCMCIA && !APNE
+	select PCCARD_IODYN
+	help
+	  Say Y here to include support for PCMCIA host bridges that
+	  are register compatible with Commodore's Gayle. This is found on
+	  Amiga systems. If unsure, say N.
+
 endif	# PCCARD
drivers/pcmcia/Makefile +1 -0
index c9d51b150682..1ebbf6c776bb 100644
@@ -33,6 +33,7 @@ obj-$(CONFIG_OMAP_CF)				+= omap_cf.o
 obj-$(CONFIG_ELECTRA_CF)			+= electra_cf.o
 obj-$(CONFIG_PCMCIA_ALCHEMY_DEVBOARD)		+= db1xxx_ss.o
 obj-$(CONFIG_PCMCIA_MAX1600)			+= max1600.o
+obj-$(CONFIG_PCMCIA_GAYLE)			+= gayle.o
 
 sa1111_cs-y					+= sa1111_generic.o
 sa1111_cs-$(CONFIG_ASSABET_NEPONSET)		+= sa1111_neponset.o
drivers/pcmcia/gayle.c new file +306 -0

306 lines not on this page · raw message

Signature · 1 line
2.34.1
#2 lore raw
In reply to Paolo Pisati

Hi Paolo,

Quoted 4 lines · Paolo Pisati

Here is a second attempt to get the PCMCIA side of Gayle's to work: it merges my previous attempt[1], some historical (e.g. v2.6.10 era) work from Kars de Jong[2](the set_io_map() and set_mem_map() are a blatant cut&paste), and it follows the NetBSD code[3] (when it made sense) - credit goes where credit is due.

Thanks for your series!

Quoted 4 lines

And, finally, the extremly sad part:

1) i've forced the word and long I/O helpers to be little endian, and this is clearly a big no (see io_mm.h).

Can't that be avoided by setting ISA_SEX/isa_sex appropriately?

Gr{oetje,eeting}s,

Geert

Signature · 5 lines
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- [email protected]

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds
#3 lore raw
Quoted 5 lines · Geert Uytterhoeven
Quoted 3 lines

1) i've forced the word and long I/O helpers to be little endian, and this is clearly a big no (see io_mm.h).

Can't that be avoided by setting ISA_SEX/isa_sex appropriately?

Hi Geert,

i tried ISA_SEX to do the be->le flipping but it broke all drivers: it seems like isa_insw()/isa_outsw()/isa_insl()/isa_outsl() must retain the host endianess.

Looking into it.

Signature · 2 lines
bye,
p.
#4 lore raw

Hi Paolo,

Quoted 9 lines · Paolo Pisati
Quoted 4 lines · Geert Uytterhoeven
Quoted 2 lines

1) i've forced the word and long I/O helpers to be little endian, and this is clearly a big no (see io_mm.h).

Can't that be avoided by setting ISA_SEX/isa_sex appropriately?

i tried ISA_SEX to do the be->le flipping but it broke all drivers: it seems like isa_insw()/isa_outsw()/isa_insl()/isa_outsl() must retain the host endianess.

Oh, I missed you're only overriding the non-repeating variants.

So ISA I/O registers are little endian, while ISA FIFO registers are native endian, as they operate on the raw data stored somewhere else (network, disk, ...)?

Gr{oetje,eeting}s,

Geert

Signature · 5 lines
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- [email protected]

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds
#5 lore raw
Quoted 10 lines · Geert Uytterhoeven
Quoted 4 lines

i tried ISA_SEX to do the be->le flipping but it broke all drivers: it seems like isa_insw()/isa_outsw()/isa_insl()/isa_outsl() must retain the host endianess.

Oh, I missed you're only overriding the non-repeating variants.

So ISA I/O registers are little endian, while ISA FIFO registers are native endian, as they operate on the raw data stored somewhere else (network, disk, ...)?

According to NetBSD[1], word access to the PCMCIA bus should be le but Kars didn't modify isa_insw()/isa_outsw()/isa_insl()/isa_outsl(), so that was my original assumption, but i'm not 100% sure here.

1: https://github.com/NetBSD/src/commit/07364bf8b04818675b3a1a1a23424be8c90ddfa2

Signature · 2 lines
bye,
p.