mail archive of the barebox mailing list
 help / color / mirror / Atom feed
From: Jean-Christophe PLAGNIOL-VILLARD <plagnioj@jcrosoft.com>
To: Michael Olbrich <m.olbrich@pengutronix.de>
Cc: barebox@lists.infradead.org
Subject: Re: [PATCH 7/8] efi: clocksoure: use event for timer
Date: Mon, 13 Mar 2017 18:52:17 +0800	[thread overview]
Message-ID: <B918A2C9-9BD1-41C2-B996-98A0B775327F@jcrosoft.com> (raw)
In-Reply-To: <20170313103002.nu64r25rzzugmk6b@pengutronix.de>


> On 13 Mar 2017, at 6:30 PM, Michael Olbrich <m.olbrich@pengutronix.de> wrote:
> 
> On Mon, Feb 27, 2017 at 11:19:29AM +0100, Jean-Christophe PLAGNIOL-VILLARD wrote:
>> with this we can be hw generic
>> 
>> If the EFI implement timestamp protocol we could use instead of event
>> but even EDK2 Never Ever compile it for any target.
>> 
>> Signed-off-by: Jean-Christophe PLAGNIOL-VILLARD <plagnioj@jcrosoft.com>
>> ---
>> common/efi/efi.c             |   8 ++++
>> drivers/clocksource/Kconfig  |   4 ++
>> drivers/clocksource/Makefile |   1 +
>> drivers/clocksource/efi.c    | 110 +++++++++++++++++++++++++++++++++++++++++++
>> 4 files changed, 123 insertions(+)
>> create mode 100644 drivers/clocksource/efi.c
>> 
>> diff --git a/common/efi/efi.c b/common/efi/efi.c
>> index 4b589b600..05c58250f 100644
>> --- a/common/efi/efi.c
>> +++ b/common/efi/efi.c
>> @@ -353,6 +353,14 @@ efi_status_t efi_main(efi_handle_t image, efi_system_table_t *sys_table)
>> 	return EFI_SUCCESS;
>> }
>> 
>> +static int efi_core_init(void)
>> +{
>> +	struct device_d *dev = device_alloc("efi-cs", DEVICE_ID_SINGLE);
>> +
>> +	return platform_device_register(dev);
>> +}
>> +core_initcall(efi_core_init);
>> +
>> static int efi_postcore_init(void)
>> {
>> 	char *uuid;
>> diff --git a/drivers/clocksource/Kconfig b/drivers/clocksource/Kconfig
>> index 9b7f0a9d7..b12a85403 100644
>> --- a/drivers/clocksource/Kconfig
>> +++ b/drivers/clocksource/Kconfig
>> @@ -34,6 +34,10 @@ config CLOCKSOURCE_DUMMY_RATE
>> 	  The option CONFIG_CLOCKSOURCE_DUMMY_RATE is used to adjust this clocksource.
>> 	  The bigger rate valuest makes clocksource "faster".
>> 
>> +config CLOCKSOURCE_EFI
>> +	bool "Generic EFI Driver"
>> +	depends on EFI_BOOTUP
>> +
>> config CLOCKSOURCE_EFI_X86
>> 	bool "EFI X86 HW driver"
>> 	depends on EFI_BOOTUP && X86
>> diff --git a/drivers/clocksource/Makefile b/drivers/clocksource/Makefile
>> index 1fd18296e..f69b33d0b 100644
>> --- a/drivers/clocksource/Makefile
>> +++ b/drivers/clocksource/Makefile
>> @@ -3,6 +3,7 @@ obj-$(CONFIG_ARM_SMP_TWD) += arm_smp_twd.o
>> obj-$(CONFIG_CLOCKSOURCE_BCM283X) += bcm2835.o
>> obj-$(CONFIG_CLOCKSOURCE_CLPS711X) += clps711x.o
>> obj-$(CONFIG_CLOCKSOURCE_DIGIC)   += digic.o
>> +obj-$(CONFIG_CLOCKSOURCE_EFI) += efi.o
>> obj-$(CONFIG_CLOCKSOURCE_EFI_X86) += efi_x86.o
>> obj-$(CONFIG_CLOCKSOURCE_MVEBU)   += mvebu.o
>> obj-$(CONFIG_CLOCKSOURCE_NOMADIK) += nomadik.o
>> diff --git a/drivers/clocksource/efi.c b/drivers/clocksource/efi.c
>> new file mode 100644
>> index 000000000..df65dd86c
>> --- /dev/null
>> +++ b/drivers/clocksource/efi.c
>> @@ -0,0 +1,110 @@
>> +/*
>> + * Copyright (C) 2012 Jean-Christophe PLAGNIOL-VILLARD <plagnio@jcrosoft.com>
>> + *
>> + * Under GPL v2
>> + */
>> +#include <common.h>
>> +#include <init.h>
>> +#include <driver.h>
>> +#include <clock.h>
>> +#include <efi.h>
>> +#include <efi/efi.h>
>> +#include <efi/efi-device.h>
>> +#include <linux/err.h>
>> +
>> +static uint64_t ticks = 1;
>> +static void *efi_cs_evt;
>> +
>> +static uint64_t efi_cs_read(void)
>> +{
>> +	return ticks;
>> +}
>> +
>> +static void efi_cs_inc(void *event, void *ctx)
>> +{
>> +	ticks++;
>> +}
>> +
>> +/* count ticks during a 1dms */
>> +static uint64_t ticks_freq(void)
>> +{
>> +	uint64_t ticks_start, ticks_end;
>> +
>> +	ticks_start = ticks;
>> +	BS->stall(1000);
>> +	ticks_end = ticks;
>> +
>> +	return (ticks_end - ticks_start) * 1000;
>> +}
>> +
>> +/* count ticks during a 20ms delay as on qemu x86_64 the max is 100Hz */
>> +static uint64_t ticks_freq_x86(void)
>> +{
>> +	uint64_t ticks_start, ticks_end;
>> +
>> +	ticks_start = ticks;
>> +	BS->stall(20 * 1000);
>> +	ticks_end = ticks;
>> +
>> +	return (ticks_end - ticks_start) * 50;
>> +}
>> +
>> +static int efi_cs_init(struct clocksource *cs)
>> +{
>> +	efi_status_t efiret;
>> +	uint64_t freq;
>> +
>> +	efiret = BS->create_event(EFI_EVT_TIMER | EFI_EVT_NOTIFY_SIGNAL,
>> +			EFI_TPL_CALLBACK, efi_cs_inc, NULL, &efi_cs_evt);
>> +
>> +	if (EFI_ERROR(efiret))
>> +		return -efi_errno(efiret);
>> +
>> +	efiret = BS->set_timer(efi_cs_evt, EFI_TIMER_PERIODIC, 10);
>> +	if (EFI_ERROR(efiret)) {
>> +		BS->close_event(efi_cs_evt);
>> +		return -efi_errno(efiret);
>> +	}
>> +
>> +	freq = 1000 * 1000;
>> +	if (ticks_freq() < 800 * 1000) {
>> +		uint64_t nb_100ns;
>> +
>> +		freq = ticks_freq_x86();
> 
> This needs a sanity ckeck. On one hardware that I have here 'freq' is often
> (but not always) zero.
> 

Really :(

please send a patch

FYI on x86 this generic Timer will replace at runtime by the efi_x86 that use HW IP
directly

Best Regards,
J.
_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox

  reply	other threads:[~2017-03-13 10:52 UTC|newest]

Thread overview: 17+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2017-02-27 10:14 [PATCH 0/8] add generic EFI timer Jean-Christophe PLAGNIOL-VILLARD
2017-02-27 10:19 ` [PATCH 1/8] efi: add prototype and definition for creating and closing event Jean-Christophe PLAGNIOL-VILLARD
2017-02-27 10:19   ` [PATCH 2/8] efi: add prototype and definition for setting timer Jean-Christophe PLAGNIOL-VILLARD
2017-02-27 10:19   ` [PATCH 3/8] efi: move LoaderTimeInitUSec and LoaderDevicePartUUID to postcore initcall Jean-Christophe PLAGNIOL-VILLARD
2017-02-27 10:19   ` [PATCH 4/8] efi: move x86 clocksource init at core initcall level Jean-Christophe PLAGNIOL-VILLARD
2017-02-27 10:19   ` [PATCH 5/8] clocksource: allow to have multiple device from clock source Jean-Christophe PLAGNIOL-VILLARD
2017-02-28  7:09     ` Sascha Hauer
2017-03-01 15:52       ` Jean-Christophe PLAGNIOL-VILLARD
2017-03-02  7:36         ` Sascha Hauer
2017-02-27 10:19   ` [PATCH 6/8] efi: move x86 clocksource to device/driver Jean-Christophe PLAGNIOL-VILLARD
2017-02-27 10:19   ` [PATCH 7/8] efi: clocksoure: use event for timer Jean-Christophe PLAGNIOL-VILLARD
2017-03-13 10:30     ` Michael Olbrich
2017-03-13 10:52       ` Jean-Christophe PLAGNIOL-VILLARD [this message]
2017-03-13 13:16         ` [PATCH] fixup! efi: clocksoure: add EFI event timer Michael Olbrich
2017-03-13 15:55           ` Jean-Christophe PLAGNIOL-VILLARD
2017-03-14  7:16           ` Sascha Hauer
2017-02-27 10:19   ` [PATCH 8/8] efi: add veriable to report secure boot support and status Jean-Christophe PLAGNIOL-VILLARD

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=B918A2C9-9BD1-41C2-B996-98A0B775327F@jcrosoft.com \
    --to=plagnioj@jcrosoft.com \
    --cc=barebox@lists.infradead.org \
    --cc=m.olbrich@pengutronix.de \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox