Geert,
On Tue, Dec 3, 2013 at 12:07 AM, Geert Uytterhoeven
wrote:
> Going back into time...
>
> On Mon, 27 Apr 2009, Geert Uytterhoeven wrote:
>> 4. CONFIG_RTC_CLASS=y, CONFIG_RTC_DRV_CMOS=[my], /dev/rtc is dynamic
>> While rtc-cmos can be enabled on Atari (it was derived from
>> rtc.c), it won't work yet,
>> as the necessary rtc_cmos platform device is not yet created on Atari.
>
> I gave it a try, cfr. "[PATCH 1/2] [RFC] m68k/atari: Create rtc_cmos platform
> device" below.
>
> Unfortunately it didn't work well, cfr. "[PATCH 2/2] [RFC] rtc: Disable
> RTC_DRV_CMOS on Atari" below, while rtc-generic works fine for the RTC part,
> and the separate nvram driver works for the NVRAM part.
>
> So unless someone has a good reason for using rtc_cmos (and some patches
> to fix its issues), I'll go for [PATCH 2/2].
The generic CMOS driver will have the TT_MFP_RTC interrupt available
only on a TT - maybe that's why it won't work on ARAnyM?
Nothing wrong with applying the second part of 1/2 (renaming the
RTC_PORT macro to ATARI_RTC_PORT) anyway, though.
Cheers,
Michael
> Thanks for your feedback!
>
> From 454783a49347468b5ddf5a01e00aa5b1b66d2480 Mon Sep 17 00:00:00 2001
> From: Geert Uytterhoeven
> Date: Mon, 2 Dec 2013 11:08:13 +0100
> Subject: [PATCH 1/2] [RFC] m68k/atari: Create rtc_cmos platform device
>
> Create an "rtc_cmos" platform device, so rtc-cmos can be used on Atari.
> Its IRQ is available on systems with a TT_MFP only.
>
> This requires renaming RTC_PORT(), as rtc-cmos uses the presence of this
> macro to enable support for the second NVRAM bank, which Atari doesn't
> have ("Unable to handle kernel access at virtual address 00ff8965").
>
> Signed-off-by: Geert Uytterhoeven
> ---
> arch/m68k/atari/config.c| 26 ++
> arch/m68k/include/asm/mc146818rtc.h | 10 +-
> 2 files changed, 31 insertions(+), 5 deletions(-)
>
> diff --git a/arch/m68k/atari/config.c b/arch/m68k/atari/config.c
> index 01a62161b08a..59704b127146 100644
> --- a/arch/m68k/atari/config.c
> +++ b/arch/m68k/atari/config.c
> @@ -660,6 +660,26 @@ static void atari_get_hardware_list(struct seq_file *m)
> ATARIHW_ANNOUNCE(DSP56K, "DSP56001 processor");
> }
>
> +static struct resource rtc_resources[] = {
> + [0] = {
> + .start = TT_RTC_BAS,
> + .end= TT_RTC_BAS + 2 * 2 - 1,
> + .flags = IORESOURCE_IO,
> + },
> + [1] = {
> + .start = IRQ_TT_MFP_RTC,
> + .end= IRQ_TT_MFP_RTC,
> + .flags = IORESOURCE_IRQ,
> + }
> +};
> +
> +static struct platform_device rtc_device = {
> + .name = "rtc_cmos",
> + .id = -1,
> + .resource = rtc_resources,
> + .num_resources = 1 /* Default to no IRQ */
> +};
> +
> /*
> * MSch: initial platform device support for Atari,
> * required for EtherNAT/EtherNEC/NetUSBee drivers
> @@ -865,6 +885,12 @@ int __init atari_platform_init(void)
> if (!MACH_IS_ATARI)
> return -ENODEV;
>
> + if (ATARIHW_PRESENT(TT_MFP)) {
> + /* Enable IRQ resource */
> + rtc_device.num_resources = 2;
> + }
> + platform_device_register(&rtc_device);
> +
> #ifdef CONFIG_ATARI_ETHERNAT
> {
> unsigned char *enatc_virt;
> diff --git a/arch/m68k/include/asm/mc146818rtc.h
> b/arch/m68k/include/asm/mc146818rtc.h
> index 9f70a01f73dc..05b43bf5cdf3 100644
> --- a/arch/m68k/include/asm/mc146818rtc.h
> +++ b/arch/m68k/include/asm/mc146818rtc.h
> @@ -10,16 +10,16 @@
>
> #include
>
> -#define RTC_PORT(x)(TT_RTC_BAS + 2*(x))
> +#define ATARI_RTC_PORT(x) (TT_RTC_BAS + 2*(x))
> #define RTC_ALWAYS_BCD 0
>
> #define CMOS_READ(addr) ({ \
> -atari_outb_p((addr),RTC_PORT(0)); \
> -atari_inb_p(RTC_PORT(1)); \
> +atari_outb_p((addr), ATARI_RTC_PORT(0)); \
> +atari_inb_p(ATARI_RTC_PORT(1)); \
> })
> #define CMOS_WRITE(val, addr) ({ \
> -atari_outb_p((addr),RTC_PORT(0)); \
> -atari_outb_p((val),RTC_PORT(1)); \
> +atari_outb_p((addr), ATARI_RTC_PORT(0)); \
> +atari_outb_p((val), ATARI_RTC_PORT(1)); \
> })
> #endif /* CONFIG_ATARI */
>
> --
> 1.7.9.5
>
> From 6c81cebe1f9bcf5fecae689235e86b42c9f8e28c Mon Sep 17 00:00:00 2001
> From: Geert Uytterhoeven
> Date: Mon, 2 Dec 2013 11:12:23 +0100
> Subject: [PATCH 2/2] [RFC] rtc: Disable RTC_DRV_CMOS on Atari
>
> On ARAnyM (emulating an Atari Falcon, which doesn't have an RTC IRQ),
> rtc-cmos doesn't work well:
> - The date is of by 32 years (2045 instead of 2013):
>
> rtc_cmos rtc_cmos: setting system clock to 2045-12-02 10:56:17 UTC
> (2395824977)
> - The hwclock utility doesn't work:
>
> hwclock: ioctl() to /dev/rtc to turn on update interrupts failed
> unexpectedly, errno=5: Input/output error.
>
> rtc-generic works fine for the RTC part, and nvram works for the NVRAM part,
> though.
>
> Signed-off-by: Geert Uytterhoeven
> ---
> d
