linux-acpi.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Andrew Morton <akpm@linux-foundation.org>
To: Len Brown <lenb@kernel.org>
Cc: linux-acpi@vger.kernel.org,
	Vladimir Lebedev <vladimir.p.lebedev@intel.com>,
	Len Brown <len.brown@intel.com>
Subject: Re: [PATCH 05/22] ACPI: Cache battery status instead of re-evaluating AML
Date: Thu, 19 Apr 2007 11:23:13 -0700	[thread overview]
Message-ID: <20070419112313.24e43271.akpm@linux-foundation.org> (raw)
In-Reply-To: <11734992652665-git-send-email-lenb@kernel.org>

On Fri,  9 Mar 2007 23:00:42 -0500
Len Brown <lenb@kernel.org> wrote:

> From: Vladimir Lebedev <vladimir.p.lebedev@intel.com>
> 
> /proc exports _BST in a single file, and _BST is re-evaulated
> whenever that file is read.
> 
> Sometimes user-space reads this file frequently, and on some
> systems _BST takes a long time to evaluate due to a slow EC.
> 
> Further, when we move to sysfs, the values returned from _BST
> will be in multiple files, and evaluating _BST for each
> file read would make matters worse.
> 
> Here code is added to support caching the results of _BST.
> A new module parameter "update_time" tells how many seconds the
> cached _BST should be used before it is re-evaluated.
> Currently, update_time defaults to 0, and so the
> existing behaviour of re-evaluating on each read retained.
> 
> ...
>  
> +#define acpi_battery_present(battery) battery->device->status.battery_present
> +#define acpi_battery_present_prev(battery) battery->flags.battery_present_prev
> +#define acpi_battery_alarm_present(battery) battery->flags.alarm_present
> +#define acpi_battery_init_update_flag(battery) battery->flags.init_update
> +#define acpi_battery_info_update_flag(battery) battery->flags.info_update
> +#define acpi_battery_state_update_flag(battery) battery->flags.state_update
> +#define acpi_battery_alarm_update_flag(battery) battery->flags.alarm_update
> +#define acpi_battery_power_units(battery) battery->flags.power_unit ? \
> +		ACPI_BATTERY_UNITS_AMPS : ACPI_BATTERY_UNITS_WATTS
> +#define acpi_battery_handle(battery) battery->device->handle
> +#define acpi_battery_inserted(battery) (!acpi_battery_present_prev(battery) & acpi_battery_present(battery))
> +#define acpi_battery_removed(battery) (acpi_battery_present_prev(battery) & !acpi_battery_present(battery))
> +#define acpi_battery_bid(battery) acpi_device_bid(battery->device)
> +#define acpi_battery_status_str(battery) acpi_battery_present(battery) ? "present" : "absent"

These macros are

a) insufficiently parenthesised: try evaluating

	acpi_battery_present(battery + 1);

b) ugly.  It's a real bad sign when we see

	acpi_battery_init_update_flag(battery) = 1;

   which simply isn't C.

If we really insist on this obfuscation layer then it would be much better
to do

static inline int acpi_battery_init_update_flag(struct acpi_battery *battery)
{
	return battery->flags.init_update;
}

static inline void
acpi_battery_set_init_update_flag(struct acpi_battery *battery, int value)
{
	battery->flags.init_update = value;
}

> +static void acpi_battery_mutex_lock(struct acpi_battery *battery)
> +{
> +	mutex_lock(&battery->mutex);
> +}
> +
> +static void acpi_battery_mutex_unlock(struct acpi_battery *battery)
> +{
> +	mutex_unlock(&battery->mutex);
> +}

These wrappers don't add any value and should be removed.

> +static void acpi_battery_check_result(struct acpi_battery *battery, int result)
> +{
> +	if (!battery)
> +		return;
> +
> +	if (result) {
> +		acpi_battery_init_update_flag(battery) = 1;
> +	}
> +}

We prefer not to have braces around a single statement like this.

> +static int acpi_battery_extract_package(struct acpi_battery *battery,
> +					union acpi_object *package,
> +					struct acpi_buffer *format,
> +					struct acpi_buffer *data,
> +					char *package_name)
> +{
> +	acpi_status status = AE_OK;
> +	struct acpi_buffer data_null = { 0, NULL };
> +
> +	status = acpi_extract_package(package, format, &data_null);
> +	if (status != AE_BUFFER_OVERFLOW) {
> +		ACPI_EXCEPTION((AE_INFO, status, "Extracting size %s",
> +				package_name));
> +		return -ENODEV;
> +	}
> +
> +	if (data_null.length != data->length) {
> +		if (data->pointer) {
> +			kfree(data->pointer);
> +		}

kfree(NULL) is legal and preferred kernel style is to not duplicate the
test fro NULL in the kfree() caller like this (there are multiple instances
of this in this patch).

> +		data->pointer = kzalloc(data_null.length, GFP_KERNEL);
> +		if (!data->pointer) {
> +			ACPI_EXCEPTION((AE_INFO, AE_NO_MEMORY, "kzalloc()"));
> +			return -ENOMEM;
> +		}
> +		data->length = data_null.length;
> +	}
> +
> +	status = acpi_extract_package(package, format, data);
> +	if (ACPI_FAILURE(status)) {
> +		ACPI_EXCEPTION((AE_INFO, status, "Extracting %s",
> +				package_name));
> +		return -ENODEV;
> +	}
> +
> +	return 0;
> +}
> +
>
> ...
>
> +static int acpi_battery_read_info(struct seq_file *seq, void *offset)
> +{
> +	struct acpi_battery *battery = seq->private;
> +	int result = 0;
> +	int update_result = ACPI_BATTERY_NONE_UPDATE;
> +	int update = 0;
> +
> +	acpi_battery_mutex_lock(battery);
> +
> +	update = (get_seconds() - battery->info_update_time >= update_time);
> +	update = (update | acpi_battery_info_update_flag(battery));
> +
> +	result = acpi_battery_update(battery, update, &update_result);
> +	if (result)
> +		goto end;
> +
> +	/* Battery Info (_BIF) */
> +
> +	if (update_result == ACPI_BATTERY_EASY_UPDATE) {
> +		result = acpi_battery_get_info(battery);
> +		if (result)
> +			goto end;
> +	}
> +
> +      end:
> +
> +	result = acpi_battery_read_info_print(seq, result);
> +
> +	acpi_battery_check_result(battery, result);
> +
> +	acpi_battery_info_update_flag(battery) = result;
> +
> +	acpi_battery_mutex_unlock(battery);
> +
> +	return result;
>  }

We have several copies of this function, each with tiny variations:

> +static int acpi_battery_read_state(struct seq_file *seq, void *offset)
> +{
> +	struct acpi_battery *battery = seq->private;
> +	int result = 0;
> +	int update_result = ACPI_BATTERY_NONE_UPDATE;
> +	int update = 0;
> +
> +	acpi_battery_mutex_lock(battery);
> +
> +	update = (get_seconds() - battery->state_update_time >= update_time);
> +	update = (update | acpi_battery_state_update_flag(battery));
> +
> +	result = acpi_battery_update(battery, update, &update_result);
> +	if (result)
> +		goto end;
> +
> +	/* Battery State (_BST) */
> +
> +	if (update_result == ACPI_BATTERY_EASY_UPDATE) {
> +		result = acpi_battery_get_state(battery);
> +		if (result)
> +			goto end;
> +	}
> +
> +      end:
> +
> +	result = acpi_battery_read_state_print(seq, result);
> +
> +	acpi_battery_check_result(battery, result);
> +
> +	acpi_battery_state_update_flag(battery) = result;
> +
> +	acpi_battery_mutex_unlock(battery);
> +
> +	return result;
>  }

Can we not find a way of consolidating all this duplication?

> +static int acpi_battery_read_alarm(struct seq_file *seq, void *offset)
> +{
> +	struct acpi_battery *battery = seq->private;
> +	int result = 0;
> +	int update_result = ACPI_BATTERY_NONE_UPDATE;
> +	int update = 0;
> +
> +	acpi_battery_mutex_lock(battery);
> +
> +	update = (get_seconds() - battery->alarm_update_time >= update_time);
> +	update = (update | acpi_battery_alarm_update_flag(battery));
> +
> +	result = acpi_battery_update(battery, update, &update_result);
> +	if (result)
> +		goto end;
> +
> +	/* Battery Alarm */
> +
> +	if (update_result == ACPI_BATTERY_EASY_UPDATE) {
> +		result = acpi_battery_get_alarm(battery);
> +		if (result)
> +			goto end;
> +	}
> +
> +      end:
> +
> +	result = acpi_battery_read_alarm_print(seq, result);
> +
> +	acpi_battery_check_result(battery, result);
> +
> +	acpi_battery_alarm_update_flag(battery) = result;
> +
> +	acpi_battery_mutex_unlock(battery);
> +
> +	return result;
>  }

More here.



  reply	other threads:[~2007-04-19 18:34 UTC|newest]

Thread overview: 31+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2007-03-10  4:00 ACPI patches in test for Linux-2.6.22 Len Brown
     [not found] ` <941b1971a405f95537bc6f9a738e737eb2186e45.1173498647.git.len.brown@intel.com>
2007-03-10  4:00   ` [PATCH 01/22] cpuidle take2: Core cpuidle infrastructure Len Brown
2007-03-15 12:31     ` Andi Kleen
2007-03-15 16:47       ` Len Brown
2007-03-15 13:13     ` Richard Hughes
2007-03-15 16:38       ` Len Brown
2007-03-15 16:42         ` Richard Hughes
     [not found]   ` <92648781fb5f44cee584da36eb569e88dddc4b8f.1173498651.git.len.brown@intel.com>
2007-03-10  4:00     ` [PATCH 02/22] cpuidle take2: Hookup ACPI C-states driver with cpuidle Len Brown
     [not found]   ` <c5bb38e598e68202e0d6f08b3fe0f30f12999357.1173498651.git.len.brown@intel.com>
2007-03-10  4:00     ` [PATCH 03/22] cpuidle take2: Basic documentation for cpuidle Len Brown
     [not found]   ` <a1f0eff21edac1bd87e397f56c4258b9611b5a50.1173498652.git.len.brown@intel.com>
2007-03-10  4:00     ` [PATCH 04/22] ACPI: battery: make internal names consistent with battery "state" Len Brown
     [not found]   ` <b6ce4083ed8e2a01a3a59301eabe0fc1e68a8a84.1173498652.git.len.brown@intel.com>
2007-03-10  4:00     ` [PATCH 05/22] ACPI: Cache battery status instead of re-evaluating AML Len Brown
2007-04-19 18:23       ` Andrew Morton [this message]
     [not found]   ` <9ea7d57576f40c6af03c8c9fa7a069f2222b498b.1173498653.git.len.brown@intel.com>
2007-03-10  4:00     ` [PATCH 06/22] ACPI: battery: Lindent Len Brown
     [not found]   ` <d98b2065c5ac31a35cbdfe9b47d421352f96fc3f.1173498653.git.len.brown@intel.com>
2007-03-10  4:00     ` [PATCH 07/22] cpuidle: build fix for !CPU_IDLE Len Brown
     [not found]   ` <bc17374e69a621c9b7f6f9624c9248f3b8e8503d.1173498653.git.len.brown@intel.com>
2007-03-10  4:00     ` [PATCH 08/22] cpu_idle: fix build break Len Brown
     [not found]   ` <3d869f55bda8b0b6f0cb55f9a85a6f855a016092.1173498654.git.len.brown@intel.com>
2007-03-10  4:00     ` [PATCH 09/22] cpuidle: make code static Len Brown
     [not found]   ` <b51f201f9a9af932c723492551f9e688d34cc468.1173498654.git.len.brown@intel.com>
2007-03-10  4:00     ` [PATCH 10/22] cpuidle: ladder does not depend on ACPI Len Brown
     [not found]   ` <e4382a15d138ea6c0d4277967e646216ce57040f.1173498654.git.len.brown@intel.com>
2007-03-10  4:00     ` [PATCH 12/22] ACPI: EC: Make EC to initialize first in ACPI Len Brown
     [not found]   ` <94418415a62ec4c88109385d8663e56aac54dfd6.1173498656.git.len.brown@intel.com>
2007-03-10  4:00     ` [PATCH 13/22] ACPI: EC: "Fake ECDT" workaround is not needed any longer Len Brown
     [not found]   ` <50a76c6beb6e41664f44239d866240952f56de78.1173498656.git.len.brown@intel.com>
2007-03-10  4:00     ` [PATCH 14/22] ACPI: EC: enable burst functionality in EC Len Brown
     [not found]   ` <388ea1ea778138553bc42b8d4aa26b9221faee23.1173498656.git.len.brown@intel.com>
2007-03-10  4:00     ` [PATCH 15/22] ACPI: EC: Remove casts to/from void* from ec.c Len Brown
     [not found]   ` <87340f2d12cb87a5a1e6cebe714cf61cc501b588.1173498656.git.len.brown@intel.com>
2007-03-10  4:00     ` [PATCH 16/22] ACPI: EC: Put install handlers into separate function Len Brown
     [not found]   ` <fb7d8f568d5bbb28d535bb68c0f287246d6a5fbf.1173498657.git.len.brown@intel.com>
2007-03-10  4:00     ` [PATCH 17/22] ACPI: EC: Clean ECDT and namespace parsing Len Brown
     [not found]   ` <105cfb8936ce350c426b9b2a42e52cd3a625d102.1173498657.git.len.brown@intel.com>
2007-03-10  4:00     ` [PATCH 18/22] ACPI: EC: Rename ec_ecdt to more informative boot_ec Len Brown
     [not found]   ` <4f53bd846f964c9c3c590b3bc8eb2bfe1d392b5d.1173498657.git.len.brown@intel.com>
2007-03-10  4:00     ` [PATCH 19/22] ACPI: EC: first_ec is better to be acpi_ec than acpi_device Len Brown
     [not found]   ` <2f11ffb10877e93e24ae042649666b78dd9985e6.1173498657.git.len.brown@intel.com>
2007-03-10  4:00     ` [PATCH 20/22] ACPI: EC: Cleanup of EC initialization Len Brown
     [not found]   ` <5923a5eb97e66a7678742b84352b82597673c766.1173498658.git.len.brown@intel.com>
2007-03-10  4:00     ` [PATCH 21/22] ACPI: EC: Block queries until EC is fully initialized Len Brown
     [not found]   ` <705ff592931e6738926fd6d596461010a1e8977d.1173498654.git.len.brown@intel.com>
2007-03-10  4:00     ` [PATCH 11/22] ACPI: EC: Don't use Global Lock if not asked to do so Len Brown
2007-03-10 14:12     ` Sanjoy Mahajan
2007-03-16 17:58 ` ACPI patches in test for Linux-2.6.22 Sergio Monteiro Basto
2007-03-16 18:03   ` Alexey Starikovskiy

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=20070419112313.24e43271.akpm@linux-foundation.org \
    --to=akpm@linux-foundation.org \
    --cc=len.brown@intel.com \
    --cc=lenb@kernel.org \
    --cc=linux-acpi@vger.kernel.org \
    --cc=vladimir.p.lebedev@intel.com \
    /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;
as well as URLs for NNTP newsgroup(s).