Re: HPET : Legacy Route replacement patch.

From: Randy Dunlap
Date: Mon Oct 23 2006 - 13:18:59 EST

Next message: Jiri Slaby: "Re: 2.6.19-rc2-git7 shutdown problem"
Previous message: Franck Bui-Huu: "Re: [PATCH 2.6.19-rc1 full] drivers: add LCD support"
In reply to: Om Narasimhan: "HPET : Legacy Route replacement patch."
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

On Mon, 23 Oct 2006 01:55:02 -0700 Om Narasimhan wrote:

> HI,
> This patch enables the legacy routing replacement capabilities of the HPET.
>
> HPET spec is available at www.intel.com/technology/architecture/hpetspec.htm
>
> Background : HPET is capable of replacing the IRQ0 (connected INT0
> PIN) timer interrupt. The capability register (at offset 0 of HPET
> base address) has a bit specifying if HPET chip is capbale of doing
> this. OS can read the bit either from HW or ACPI table. (HPET ACPI
> description table -> Event Timer block -> bit 15, page 30 of HPET
> spec). Ideally, I think BIOS should set the ACPI table than letting
> the OS read H/W, which gives the BIOS a way to configure either legacy
> or Legacy replacement modes.
>
> But a number of bioses (both phoenix and AMI) are not working as
> expected. So I have provided a command line parameter which overrides
> the BIOS ACPI entry. So, irrespctive of the BIOS' HPET ACPI Descriptor
> table settings, if the parameter hpet_lrr=[0,1] is specified, it takes
> precedence.

The "hpet_lrr" parameter needs to be documented in
Documentation/kernel-parameters.txt.

> It seems, some other chipsets (CK-804) has a timing problem, as a
> result of which the machine hangs. In this case, I have passed the
> parameter lpj=<approx bogomips value * 500> which solves this issue. I
> am investigating into this further and would try fixing it.

so what machines (BIOSen) do HPET legacy route replacement
correctly? (as far as you know)

> Even though compile tested, on 2.6.19-XX I have not tested this patch.
> I ran this on 2.6.16-rc5, which is the latest version that runs on my
> machine (sunfire 4600).
>
> Patch here...
>
> Signed-off-by: Om Narasimhan <om.turyx@xxxxxxxxx>
>
> ----
>
> arch/i386/kernel/acpi/boot.c | 13 +++++++++++++
> arch/i386/kernel/time_hpet.c | 3 ++-
> arch/x86_64/kernel/time.c | 15 +++++++++++++--
> include/asm-x86_64/hpet.h | 1 +
> include/linux/acpi.h | 1 +
> 5 files changed, 30 insertions(+), 3 deletions(-)
> arch/i386/kernel/acpi/boot.c | 13 +++++++++++++
> arch/i386/kernel/time_hpet.c | 3 ++-
> arch/x86_64/kernel/time.c | 15 +++++++++++++--
> include/asm-x86_64/hpet.h | 1 +
> include/linux/acpi.h | 1 +
> 5 files changed, 30 insertions(+), 3 deletions(-)
>
> diff --git a/arch/i386/kernel/acpi/boot.c b/arch/i386/kernel/acpi/boot.c
> index ab974ff..228d2b5 100644
> --- a/arch/i386/kernel/acpi/boot.c
> +++ b/arch/i386/kernel/acpi/boot.c
> @@ -82,6 +82,15 @@ EXPORtested T_SYMBOL(acpi_strict);
> acpi_interrupt_flags acpi_sci_flags __initdata;
> int acpi_sci_override_gsi __initdata;
> int acpi_skip_timer_override __initdata;
> +int acpi_hpet_lrr; /* the HPET Legacy routing replacement option
> passed through ACPI Table */
> +int hpet_lrr_force; /* cmdlinetested opt. for faulty bioses not
> setting ACPI HPET entry right */

Lines too long, please limit to < 80 characters each.

> +static int hpet_lrr_setup (char *str)
> +{
> + get_option(&str,&hpet_lrr_force);

space after comma, please.

> + return 1;
> +}
> +__setup ("hpet_lrr=", hpet_lrr_setup);
>
> #ifdef CONFIG_X86_LOCAL_APIC
> static u64 acpi_lapic_addr __initdata = APIC_DEFAULT_PHYS_BASE;
> @@ -669,6 +678,10 @@ #define HPET_RESOURCE_NAME_SIZE 9
> "HPET %u", hpet_tbl->number);
> hpet_res->end = (1 * 1024) - 1;
> }
> + acpi_hpet_lrr = (hpet_tbl->id & ACPI_HPET_LRR_CAP) ? 1 : 0;
> + /* Print a message to warn about the bios BUG if HPET ACPI is not
> correctly populated */
> + printk(KERN_INFO PREFIX "HPET id: %#x. ACPI LRR bit %s SET\n",
> + hpet_tbl->id, acpi_hpet_lrr ? "": "NOT");
>
> #ifdef CONFIG_X86_64
> vxtime.hpet_address = hpet_tbl->addr.addrl |

> diff --git a/arch/x86_64/kernel/time.c b/arch/x86_64/kernel/time.c
> index 1ba5a44..3417c0f 100644
> --- a/arch/x86_64/kernel/time.c
> +++ b/arch/x86_64/kernel/time.c
> @@ -48,6 +48,8 @@ static void cpufreq_delayed_get(void);
> #endif
> extern void i8254_timer_resume(void);
> extern int using_apic_timer;
> +extern int acpi_hpet_lrr;
> +extern int hpet_lrr_force;

All of these externs (not just the added ones) should be in some
header file which is #included here.

> static char *timename = NULL;
>

> @@ -924,7 +934,8 @@ #endif
> vxtime.tsc_quot = (USEC_PER_MSEC << US_SCALE) / cpu_khz;
> vxtime.last_tsc = get_cycles_sync();
> set_cyc2ns_scale(cpu_khz);
> - setup_irq(0, &irq0);
> + printk (KERN_WARNING PREFIX "Registering Timer IRQ = %d\n", timer_irq);

No space between function name and '('.

> + setup_irq(timer_irq, &irq0);
> hotcpu_notifier(time_cpu_notifier, 0);
> time_cpu_notifier(NULL, CPU_ONLINE, (void *)(long)smp_processor_id());
>

> diff --git a/arch/i386/kernel/acpi/boot.c b/arch/i386/kernel/acpi/boot.c
> index ab974ff..228d2b5 100644
> --- a/arch/i386/kernel/acpi/boot.c
> +++ b/arch/i386/kernel/acpi/boot.c
> @@ -82,6 +82,15 @@ EXPORtested T_SYMBOL(acpi_strict);
> acpi_interrupt_flags acpi_sci_flags __initdata;
> int acpi_sci_override_gsi __initdata;
> int acpi_skip_timer_override __initdata;
> +int acpi_hpet_lrr; /* the HPET Legacy routing replacement option
> passed through ACPI Table */
> +int hpet_lrr_force; /* cmdlinetested opt. for faulty bioses not
> setting ACPI HPET entry right */

Lines too long...

> +static int hpet_lrr_setup (char *str)
> +{
> + get_option(&str,&hpet_lrr_force);
> + return 1;
> +}
> +__setup ("hpet_lrr=", hpet_lrr_setup);
>
> #ifdef CONFIG_X86_LOCAL_APIC
> static u64 acpi_lapic_addr __initdata = APIC_DEFAULT_PHYS_BASE;
> @@ -669,6 +678,10 @@ #define HPET_RESOURCE_NAME_SIZE 9
> "HPET %u", hpet_tbl->number);
> hpet_res->end = (1 * 1024) - 1;
> }
> + acpi_hpet_lrr = (hpet_tbl->id & ACPI_HPET_LRR_CAP) ? 1 : 0;
> + /* Print a message to warn about the bios BUG if HPET ACPI is not
> correctly populated */

Line wrap, MUA problem? When joined into one line, the line is
too long. Please limit lines to < 80 characters each.

Several instances of this, causing "malformed patch" messages
from patch:
patch: **** malformed patch at line 58: passed through ACPI Table */

> + printk(KERN_INFO PREFIX "HPET id: %#x. ACPI LRR bit %s SET\n",
> + hpet_tbl->id, acpi_hpet_lrr ? "": "NOT");
>
> #ifdef CONFIG_X86_64
> vxtime.hpet_address = hpet_tbl->addr.addrl |

> diff --git a/arch/x86_64/kernel/time.c b/arch/x86_64/kernel/time.c
> index 1ba5a44..3417c0f 100644
> --- a/arch/x86_64/kernel/time.c
> +++ b/arch/x86_64/kernel/time.c
> @@ -48,6 +48,8 @@ static void cpufreq_delayed_get(void);
> #endif
> extern void i8254_timer_resume(void);
> extern int using_apic_timer;
> +extern int acpi_hpet_lrr;
> +extern int hpet_lrr_force;

Put externs into a header file and include that file.

> static char *timename = NULL;
>

> @@ -924,7 +934,8 @@ #endif
> vxtime.tsc_quot = (USEC_PER_MSEC << US_SCALE) / cpu_khz;
> vxtime.last_tsc = get_cycles_sync();
> set_cyc2ns_scale(cpu_khz);
> - setup_irq(0, &irq0);
> + printk (KERN_WARNING PREFIX "Registering Timer IRQ = %d\n", timer_irq);

no space between printk and (

> + setup_irq(timer_irq, &irq0);
> hotcpu_notifier(time_cpu_notifier, 0);
> time_cpu_notifier(NULL, CPU_ONLINE, (void *)(long)smp_processor_id());
>

> diff --git a/include/linux/acpi.h b/include/linux/acpi.h
> index 2b0c955..62dab08 100644
> --- a/include/linux/acpi.h arch/i386/kernel/acpi/boot.c | 13 +++++++++++++
> arch/i386/kernel/time_hpet.c | 3 ++-
> arch/x86_64/kernel/time.c | 15 +++++++++++++--
> include/asm-x86_64/hpet.h | 1 +
> include/linux/acpi.h | 1 +
> 5 files changed, 30 insertions(+), 3 deletions(-)

??? more malformed patch...

> diff --git a/arch/i386/kernel/acpi/boot.c b/arch/i386/kernel/acpi/boot.c
> index ab974ff..228d2b5 100644
> --- a/arch/i386/kernel/acpi/boot.c
> +++ b/arch/i386/kernel/acpi/boot.c
> @@ -82,6 +82,15 @@ EXPORtested T_SYMBOL(acpi_strict);
> acpi_interrupt_flags acpi_sci_flags __initdata;
> int acpi_sci_override_gsi __initdata;
> int acpi_skip_timer_override __initdata;
> +int acpi_hpet_lrr; /* the HPET Legacy routing replacement option
> passed through ACPI Table */
> +int hpet_lrr_force; /* cmdlinetested opt. for faulty bioses not
> setting ACPI HPET entry right */

Lines too long.

> +static int hpet_lrr_setup (char *str)
> +{
> + get_option(&str,&hpet_lrr_force);

space after comma, please.

> + return 1;
> +}
> +__setup ("hpet_lrr=", hpet_lrr_setup);
>
> #ifdef CONFIG_X86_LOCAL_APIC
> static u64 acpi_lapic_addr __initdata = APIC_DEFAULT_PHYS_BASE;
> @@ -669,6 +678,10 @@ #define HPET_RESOURCE_NAME_SIZE 9
> "HPET %u", hpet_tbl->number);
> hpet_res->end = (1 * 1024) - 1;
> }
> + acpi_hpet_lrr = (hpet_tbl->id & ACPI_HPET_LRR_CAP) ? 1 : 0;
> + /* Print a message to warn about the bios BUG if HPET ACPI is not
> correctly populated */
> + printk(KERN_INFO PREFIX "HPET id: %#x. ACPI LRR bit %s SET\n",
> + hpet_tbl->id, acpi_hpet_lrr ? "": "NOT");
>
> #ifdef CONFIG_X86_64
> vxtime.hpet_address = hpet_tbl->addr.addrl |

> diff --git a/arch/x86_64/kernel/time.c b/arch/x86_64/kernel/time.c
> index 1ba5a44..3417c0f 100644
> --- a/arch/x86_64/kernel/time.c
> +++ b/arch/x86_64/kernel/time.c
> @@ -48,6 +48,8 @@ static void cpufreq_delayed_get(void);
> #endif
> extern void i8254_timer_resume(void);
> extern int using_apic_timer;
> +extern int acpi_hpet_lrr;
> +extern int hpet_lrr_force;

use some header file for these.

> static char *timename = NULL;
>

> @@ -887,6 +892,7 @@ time_cpu_notifier(struct notifier_block
>
> void __init time_init(void)
> {
> + int timer_irq = 0;
> if (nohpet)
> vxtime.hpet_address = 0;
>

> diff --git a/include/linux/acpi.h b/include/linux/acpi.h
> index 2b0c955..62dab08 100644
> --- a/include/linux/acpi.h
> +++ b/include/linux/acpi.h
> @@ -220,6 +220,7 @@ enum acpi_interrupt_id {
> };
>
> #define ACPI_SPACE_MEM 0
> +#define ACPI_HPET_LRR_CAP 0x8000
>
> struct acpi_gen_regaddr {
> u8 space_id;

some kind of diff header corruption here? Missing a few lines:

> +++ b/include/linux/acpi.h
> @@ -220,6 +220,7 @@ enum acpi_interrupt_id {
> };
>
> #define ACPI_SPACE_MEM 0
> +#define ACPI_HPET_LRR_CAP 0x8000
>
> struct acpi_gen_regaddr {
> u8 space_id;
> -

---
~Randy
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

Next message: Jiri Slaby: "Re: 2.6.19-rc2-git7 shutdown problem"
Previous message: Franck Bui-Huu: "Re: [PATCH 2.6.19-rc1 full] drivers: add LCD support"
In reply to: Om Narasimhan: "HPET : Legacy Route replacement patch."
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]