All of lore.kernel.org
 help / color / mirror / Atom feed
From: "Rafael J. Wysocki" <rjw@sisk.pl>
To: linux-pm@lists.linux-foundation.org
Cc: Hidetoshi Seto <seto.hidetoshi@jp.fujitsu.com>,
	linux-pci@vger.kernel.org,
	ACPI Devel Maling List <linux-acpi@vger.kernel.org>,
	Jesse Barnes <jbarnes@virtuousgeek.org>,
	Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com>
Subject: Re: [linux-pm] [PATCH 7/8] ACPI / PCI: Do not preserve _OSC control bits returned by a query (v2)
Date: Thu, 5 Aug 2010 01:51:27 +0200	[thread overview]
Message-ID: <201008050151.27501.rjw@sisk.pl> (raw)
In-Reply-To: <201008041229.19848.rjw@sisk.pl>

On Wednesday, August 04, 2010, Rafael J. Wysocki wrote:
> On Wednesday, August 04, 2010, Hidetoshi Seto wrote:
> > (2010/08/04 14:46), Kenji Kaneshige wrote:
> > > (2010/08/04 6:02), Rafael J. Wysocki wrote:
> > >> On Tuesday, August 03, 2010, Rafael J. Wysocki wrote:
> > >>> On Tuesday, August 03, 2010, Hidetoshi Seto wrote:
> > >>>> (2010/08/03 13:52), Kenji Kaneshige wrote:
> > >>>>> (2010/08/03 6:59), Rafael J. Wysocki wrote:
> > >>>>>> @@ -434,19 +432,6 @@ acpi_status acpi_pci_osc_control_set(acp
> > >>>>>>        if ((root->osc_control_set&   control_req) == control_req)
> > >>>>>>            goto out;
> > >>>>>>
> > >>>>>> -    /* Need to query controls first before requesting them */
> > >>>>>> -    if (!root->osc_queried) {
> > >>>>>> -        status = acpi_pci_query_osc(root, root->osc_support_set, NULL);
> > >>>>>> -        if (ACPI_FAILURE(status))
> > >>>>>> -            goto out;
> > >>>>>> -    }
> > >>>>>> -    if ((root->osc_control_qry&   control_req) != control_req) {
> > >>>>>> -        printk(KERN_DEBUG
> > >>>>>> -               "Firmware did not grant requested _OSC control\n");
> > >>>>>> -        status = AE_SUPPORT;
> > >>>>>> -        goto out;
> > >>>>>> -    }
> > >>>>>
> > >>>>> I think acpi_pci_osc_control_set() still need to query before commit
> > >>>>> to ensure all the requested controls are granted to OS.
> > >>>>>
> > >>>>> So the code needs to be
> > >>>>>
> > >>>>>      status = acpi_pci_query_osc(root, root->osc_support_set,&control_req);
> > >>>>>      if (ACPI_FAILURE(status))
> > >>>>>          goto out;
> > >>>>
> > > 
> > > Sorry, that should have been
> > > 
> > >     query = control_req;
> > >     status = acpi_pci_query_osc(root, root->osc_support_set, &query);
> > >     if (ACPI_FAILURE(status))
> > >         goto out;
> > >     if ((query & control_req) != control_req) {
> > >         printk_(KERN_DEBUG
> > >             "Firmware did not grant requested _OSC control\n");
> > >         status = AE_SUPPORT;
> > >         goto out;
> > >     }
> > > 
> > > I know current pcie_port_acpi_setup() queries the requesting controls
> > > before acpi_pci_osc_control_set() and only one control is requested
> > > in the other code. However, I think acpi_pci_osc_control_set() still
> > > need to query the requested controls to ensure all the requested
> > > controls, in case someone calls this function without querying the
> > > requesting controls. In other words, I think it must be ensured that
> > > any controls are never granted to OS when acpi_pci_osc_control_set()
> > > returns error.
> > 
> > I think the following patch is what you mean.
> > 
> > And... (Continue to next post)
> > 
> > Thanks,
> > H.Seto
> 
> OK, I'll repalce my 7/8 with the patch below.

Actually, having reconsidered that, I don't think this approach is valid.

First, it has the problem that if acpi_pci_osc_control_set() returns error
code, the caller doesn't really know whether the query failed, or the final
request failed.  Arguably, it won't matter for the majority of callers, but
some of them might be interested in knowing that in principle.

Second, the callers that call acpi_pci_osc_control_query() before
acpi_pci_osc_control_set() don't need the additional query inside
of acpi_pci_osc_control_set().

Therefore I'd prefer to have two separate functions, one for querying and the
other for requesting control.  Then, we can provide a helper that calls the
both of them for the callers of acpi_pci_osc_control_set() that don't need
to call acpi_pci_osc_control_query() directly by themselves.

Given that acpi_pci_osc_control_query() is introduced by
https://patchwork.kernel.org/patch/117176/ , the helper may be implemented
like in the appended patch.

After that patch, the $subject patch can be applied without any modifications
to remove the no-longer-used fields in struct acpi_pci_root.

Thanks,
Rafael

---
From: Rafael J. Wysocki <rjw@sisk.pl>
Subject: ACPI / PCI: Introduce function for requesting _OSC controls safely

Calling raw acpi_pci_osc_control_set() is generally unsafe, because
it may return error code even if control of some requested features
have been granted by the BIOS.  For this reason the callers of
acpi_pci_osc_control_set() should call acpi_pci_osc_control_query()
before it to make sure that the BIOS is willing to grant control of
the requested features.

Introduce helper function acpi_pci_osc_control_set_safe() allowing
a caller of acpi_pci_osc_control_set() who is not interested in
the control bits returned by acpi_pci_osc_control_query() to
request control of _OSC features in a safe way.

Make acpi_get_hp_hw_control_from_firmware() use the new function.

Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
---
 drivers/acpi/pci_root.c          |   24 ++++++++++++++++++++++++
 drivers/pci/hotplug/acpi_pcihp.c |    2 +-
 include/linux/acpi.h             |    1 +
 3 files changed, 26 insertions(+), 1 deletion(-)

Index: linux-2.6/drivers/acpi/pci_root.c
===================================================================
--- linux-2.6.orig/drivers/acpi/pci_root.c
+++ linux-2.6/drivers/acpi/pci_root.c
@@ -472,6 +472,30 @@ out:
 }
 EXPORT_SYMBOL(acpi_pci_osc_control_set);
 
+/**
+ * acpi_pci_osc_control_set_safe - Query and set _OSC control bit mask.
+ * @handle: ACPI handle of a PCI root bridge (or PCIe Root Complex).
+ * @flags: Mask of _OSC bits to query and set.
+ *
+ * Check if the BIOS is willing to grant control of the features represented
+ * by @flags and request control of these features from it.
+ **/
+acpi_status acpi_pci_osc_control_set_safe(acpi_handle handle, u32 flags)
+{
+	acpi_status status;
+	u32 ctrl = flags;
+
+	status = acpi_pci_osc_control_query(handle, &flags);
+	if (ACPI_FAILURE(status))
+		return status;
+	if ((ctrl & flags) != ctrl)
+		return AE_SUPPORT;
+
+	status = acpi_pci_osc_control_set(handle, flags);
+	return status;
+}
+EXPORT_SYMBOL(acpi_pci_osc_control_set_safe);
+
 static int __devinit acpi_pci_root_add(struct acpi_device *device)
 {
 	unsigned long long segment, bus;
Index: linux-2.6/drivers/pci/hotplug/acpi_pcihp.c
===================================================================
--- linux-2.6.orig/drivers/pci/hotplug/acpi_pcihp.c
+++ linux-2.6/drivers/pci/hotplug/acpi_pcihp.c
@@ -358,7 +358,7 @@ int acpi_get_hp_hw_control_from_firmware
 		acpi_get_name(handle, ACPI_FULL_PATHNAME, &string);
 		dbg("Trying to get hotplug control for %s\n",
 				(char *)string.pointer);
-		status = acpi_pci_osc_control_set(handle, flags);
+		status = acpi_pci_osc_control_set_safe(handle, flags);
 		if (ACPI_SUCCESS(status))
 			goto got_one;
 		if (status == AE_SUPPORT)
Index: linux-2.6/include/linux/acpi.h
===================================================================
--- linux-2.6.orig/include/linux/acpi.h
+++ linux-2.6/include/linux/acpi.h
@@ -307,6 +307,7 @@ acpi_status acpi_run_osc(acpi_handle han
 
 extern acpi_status acpi_pci_osc_control_query(acpi_handle handle, u32 *mask);
 extern acpi_status acpi_pci_osc_control_set(acpi_handle handle, u32 flags);
+extern acpi_status acpi_pci_osc_control_set_safe(acpi_handle handle, u32 flags);
 extern void acpi_early_init(void);
 
 #else	/* !CONFIG_ACPI */

  reply	other threads:[~2010-08-04 23:51 UTC|newest]

Thread overview: 85+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2010-08-02 21:51 [PATCH 0/8] ACPI / PCI / PCIe: Rework _OSC handling (v2) Rafael J. Wysocki
2010-08-02 21:53 ` [PATCH 1/8] ACPI / PCI: Introduce acpi_pci_osc_control_query() Rafael J. Wysocki
2010-08-02 21:53 ` Rafael J. Wysocki
2010-08-03  4:52   ` Kenji Kaneshige
2010-08-03  4:52   ` Kenji Kaneshige
2010-08-03  9:24     ` Rafael J. Wysocki
2010-08-03  9:24     ` Rafael J. Wysocki
2010-08-03  5:04   ` Kenji Kaneshige
2010-08-03  5:04   ` Kenji Kaneshige
2010-08-03  9:27     ` Rafael J. Wysocki
2010-08-03  9:27     ` Rafael J. Wysocki
2010-08-03 20:58       ` Rafael J. Wysocki
2010-08-03 20:58       ` [linux-pm] " Rafael J. Wysocki
2010-08-04  4:28         ` Kenji Kaneshige
2010-08-04  4:28         ` [linux-pm] " Kenji Kaneshige
2010-08-04 23:37           ` Rafael J. Wysocki
2010-08-04 23:37           ` [linux-pm] " Rafael J. Wysocki
2010-08-05 23:46             ` Rafael J. Wysocki
2010-08-05 23:46             ` Rafael J. Wysocki
2010-08-02 21:54 ` [PATCH 2/8] PCI / PCIe/ AER: Introduce pci_aer_available() Rafael J. Wysocki
2010-08-03  0:46   ` Hidetoshi Seto
2010-08-03  0:46   ` Hidetoshi Seto
2010-08-03  9:28     ` Rafael J. Wysocki
2010-08-03  9:28     ` Rafael J. Wysocki
2010-08-03 20:59       ` Rafael J. Wysocki
2010-08-03 20:59       ` [linux-pm] " Rafael J. Wysocki
2010-08-03  9:41   ` Jike Song
2010-08-03  9:41   ` Jike Song
2010-08-03 19:50     ` Rafael J. Wysocki
2010-08-03 19:50     ` Rafael J. Wysocki
2010-08-02 21:54 ` Rafael J. Wysocki
2010-08-02 21:55 ` [PATCH 3/8] PCI / PCIe: Introduce commad line switch for disabling port services Rafael J. Wysocki
2010-08-02 21:55 ` Rafael J. Wysocki
2010-08-02 21:56 ` [PATCH 4/8] PCI / PCIe: Ask BIOS for control of all native services at once (v6) Rafael J. Wysocki
2010-08-02 21:56 ` Rafael J. Wysocki
2010-08-03  1:14   ` Hidetoshi Seto
2010-08-03  1:14   ` Hidetoshi Seto
2010-08-03 21:01     ` Rafael J. Wysocki
2010-08-03 21:01     ` Rafael J. Wysocki
2010-08-06  1:33   ` Hidetoshi Seto
2010-08-06  1:33   ` Hidetoshi Seto
2010-08-06 10:47     ` Rafael J. Wysocki
2010-08-02 21:57 ` [PATCH 5/8] PCI / PCIe: Disable PCIe port services during port initialization Rafael J. Wysocki
2010-08-02 21:57 ` Rafael J. Wysocki
2010-08-02 21:58 ` [PATCH 6/8] PCI / PCIe: Remove the port driver module exit routine Rafael J. Wysocki
2010-08-02 21:58 ` Rafael J. Wysocki
2010-08-02 21:59 ` [PATCH 7/8] ACPI / PCI: Do not preserve _OSC control bits returned by a query (v2) Rafael J. Wysocki
2010-08-02 21:59 ` Rafael J. Wysocki
2010-08-03  4:52   ` Kenji Kaneshige
2010-08-03  4:52   ` Kenji Kaneshige
2010-08-03  7:13     ` Hidetoshi Seto
2010-08-03  9:33       ` Rafael J. Wysocki
2010-08-03 21:02         ` Rafael J. Wysocki
2010-08-03 21:02         ` [linux-pm] " Rafael J. Wysocki
2010-08-04  5:46           ` Kenji Kaneshige
2010-08-04  5:46           ` [linux-pm] " Kenji Kaneshige
2010-08-04  8:41             ` Hidetoshi Seto
2010-08-04  9:23               ` Kenji Kaneshige
2010-08-04  9:23               ` [linux-pm] " Kenji Kaneshige
2010-08-04 10:29               ` Rafael J. Wysocki
2010-08-04 10:29               ` [linux-pm] " Rafael J. Wysocki
2010-08-04 23:51                 ` Rafael J. Wysocki [this message]
2010-08-05  3:40                   ` Hidetoshi Seto
2010-08-05 14:25                     ` Rafael J. Wysocki
2010-08-05 14:25                     ` [linux-pm] " Rafael J. Wysocki
2010-08-06  1:28                       ` Hidetoshi Seto
2010-08-06  1:28                       ` Hidetoshi Seto
2010-08-05  3:40                   ` Hidetoshi Seto
2010-08-04 23:51                 ` Rafael J. Wysocki
2010-08-04  8:41             ` Hidetoshi Seto
2010-08-04  8:43             ` Hidetoshi Seto
2010-08-04  8:43             ` [linux-pm] " Hidetoshi Seto
2010-08-04  9:39               ` Kenji Kaneshige
2010-08-04 12:15                 ` Rafael J. Wysocki
2010-08-05  3:38                   ` Hidetoshi Seto
2010-08-05  3:38                   ` [linux-pm] " Hidetoshi Seto
2010-08-04  9:39               ` Kenji Kaneshige
2010-08-04 10:38               ` [linux-pm] " Rafael J. Wysocki
2010-08-04 10:38               ` Rafael J. Wysocki
2010-08-03  9:33       ` Rafael J. Wysocki
2010-08-03  7:13     ` Hidetoshi Seto
2010-08-02 22:00 ` [PATCH 8/8] ACPI / PCI: Reorder checks in acpi_pci_osc_control_set() Rafael J. Wysocki
2010-08-02 22:00 ` Rafael J. Wysocki
2010-08-03  4:51 ` [PATCH 0/8] ACPI / PCI / PCIe: Rework _OSC handling (v2) Kenji Kaneshige
2010-08-03  4:51 ` Kenji Kaneshige

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=201008050151.27501.rjw@sisk.pl \
    --to=rjw@sisk.pl \
    --cc=jbarnes@virtuousgeek.org \
    --cc=kaneshige.kenji@jp.fujitsu.com \
    --cc=linux-acpi@vger.kernel.org \
    --cc=linux-pci@vger.kernel.org \
    --cc=linux-pm@lists.linux-foundation.org \
    --cc=seto.hidetoshi@jp.fujitsu.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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.