From: Vaibhav Gupta <vaibhavgupta40@gmail.com>
To: Sam Ravnborg <sam@ravnborg.org>
Cc: linux-fbdev@vger.kernel.org, Antonino Daplas <adaplas@gmail.com>,
Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com>,
Vaibhav Gupta <vaibhav.varodek@gmail.com>,
Florian Tobias Schandinat <FlorianSchandinat@gmx.de>,
Russell King <linux@armlinux.org.uk>,
dri-devel@lists.freedesktop.org, linux-kernel@vger.kernel.org,
Bjorn Helgaas <helgaas@kernel.org>,
linux-arm-kernel@lists.infradead.org,
Andres Salomon <dilinger@queued.net>,
Bjorn Helgaas <bhelgaas@google.com>,
Paul Mackerras <paulus@samba.org>,
linux-kernel-mentees@lists.linuxfoundation.org,
linux-geode@lists.infradead.org
Subject: Re: [Linux-kernel-mentees] [PATCH v1 01/12] fbdev: gxfb: use generic power management
Date: Mon, 10 Aug 2020 15:09:48 +0530 [thread overview]
Message-ID: <20200810093948.GB6615@gmail.com> (raw)
In-Reply-To: <20200808111746.GA24172@ravnborg.org>
On Sat, Aug 08, 2020 at 01:17:46PM +0200, Sam Ravnborg wrote:
> Hi Vaibhav
>
> On Wed, Aug 05, 2020 at 11:37:11PM +0530, Vaibhav Gupta wrote:
> > Drivers using legacy power management .suspen()/.resume() callbacks
> > have to manage PCI states and device's PM states themselves. They also
> > need to take care of standard configuration registers.
> >
> > Switch to generic power management framework using a single
> > "struct dev_pm_ops" variable
>
> "to take the unnecessary load from the driver."
> - I do not parse the above - I cannot see what load is removed.
> But the code is simpler which is fine. The drawback is that we now
> always link in the suspend_gx functions but hopefultl the linker drops
> them later.
>
> > This also avoids the need for the driver to directly call most of the PCI
> > helper functions and device power state control functions, as through
> > the generic framework PCI Core takes care of the necessary operations,
> > and drivers are required to do only device-specific jobs.
> Again, I do not see what calles are removed.
> A single check for the state is dropped - anything else?
>
Yeah, the commit messages are bit misleading, I have modified them.
> >
> > Signed-off-by: Vaibhav Gupta <vaibhavgupta40@gmail.com>
> > ---
> > drivers/video/fbdev/geode/gxfb.h | 5 ----
> > drivers/video/fbdev/geode/gxfb_core.c | 36 ++++++++++++++------------
> > drivers/video/fbdev/geode/suspend_gx.c | 4 ---
> > 3 files changed, 20 insertions(+), 25 deletions(-)
> >
> > diff --git a/drivers/video/fbdev/geode/gxfb.h b/drivers/video/fbdev/geode/gxfb.h
> > index d2e9c5c8e294..792c111c21e4 100644
> > --- a/drivers/video/fbdev/geode/gxfb.h
> > +++ b/drivers/video/fbdev/geode/gxfb.h
> > @@ -21,7 +21,6 @@ struct gxfb_par {
> > void __iomem *dc_regs;
> > void __iomem *vid_regs;
> > void __iomem *gp_regs;
> > -#ifdef CONFIG_PM
> > int powered_down;
> >
> > /* register state, for power management functionality */
> > @@ -36,7 +35,6 @@ struct gxfb_par {
> > uint64_t fp[FP_REG_COUNT];
> >
> > uint32_t pal[DC_PAL_COUNT];
> > -#endif
> > };
> >
> > unsigned int gx_frame_buffer_size(void);
> > @@ -49,11 +47,8 @@ void gx_set_dclk_frequency(struct fb_info *info);
> > void gx_configure_display(struct fb_info *info);
> > int gx_blank_display(struct fb_info *info, int blank_mode);
> >
> > -#ifdef CONFIG_PM
> > int gx_powerdown(struct fb_info *info);
> > int gx_powerup(struct fb_info *info);
> > -#endif
> > -
> >
> > /* Graphics Processor registers (table 6-23 from the data book) */
> > enum gp_registers {
> > diff --git a/drivers/video/fbdev/geode/gxfb_core.c b/drivers/video/fbdev/geode/gxfb_core.c
> > index d38a148d4746..44089b331f91 100644
> > --- a/drivers/video/fbdev/geode/gxfb_core.c
> > +++ b/drivers/video/fbdev/geode/gxfb_core.c
> > @@ -322,17 +322,14 @@ static struct fb_info *gxfb_init_fbinfo(struct device *dev)
> > return info;
> > }
> >
> > -#ifdef CONFIG_PM
> > -static int gxfb_suspend(struct pci_dev *pdev, pm_message_t state)
> > +static int __maybe_unused gxfb_suspend(struct device *dev)
> > {
> > - struct fb_info *info = pci_get_drvdata(pdev);
> > + struct fb_info *info = dev_get_drvdata(dev);
> I do not see any dev_set_drvdata() so I guess we get a NULL pointer
> here which is not intended.
> Adding a dev_set_data() to gxfb_probe() would do the trick.
>
gxfb_probe() invokes pci_set_drvdata(pdev, info) which in turn calls
dev_set_drvdata(&pdev->dev, data). Adding dev_get_drvdata() will be redundant.
> >
> > - if (state.event == PM_EVENT_SUSPEND) {
> > - console_lock();
> > - gx_powerdown(info);
> > - fb_set_suspend(info, 1);
> > - console_unlock();
> > - }
> > + console_lock();
> > + gx_powerdown(info);
> > + fb_set_suspend(info, 1);
> > + console_unlock();
> >
> > /* there's no point in setting PCI states; we emulate PCI, so
> > * we don't end up getting power savings anyways */
> > @@ -340,9 +337,9 @@ static int gxfb_suspend(struct pci_dev *pdev, pm_message_t state)
> > return 0;
> > }
> >
> > -static int gxfb_resume(struct pci_dev *pdev)
> > +static int __maybe_unused gxfb_resume(struct device *dev)
> > {
> > - struct fb_info *info = pci_get_drvdata(pdev);
> > + struct fb_info *info = dev_get_drvdata(dev);
> > int ret;
> >
> > console_lock();
> > @@ -356,7 +353,6 @@ static int gxfb_resume(struct pci_dev *pdev)
> > console_unlock();
> > return 0;
> > }
> > -#endif
> >
> > static int gxfb_probe(struct pci_dev *pdev, const struct pci_device_id *id)
> > {
> > @@ -467,15 +463,23 @@ static const struct pci_device_id gxfb_id_table[] = {
> >
> > MODULE_DEVICE_TABLE(pci, gxfb_id_table);
> >
> > +static const struct dev_pm_ops gxfb_pm_ops = {
> > +#ifdef CONFIG_PM_SLEEP
> > + .suspend = gxfb_suspend,
> > + .resume = gxfb_resume,
> > + .freeze = NULL,
> > + .thaw = gxfb_resume,
> > + .poweroff = NULL,
> > + .restore = gxfb_resume,
> > +#endif
> > +};
> Can we use SET_SYSTEM_SLEEP_PM_OPS here?
> .freeze will be assigned gxfb_suspend, but gxfb_suspend will anyway be
> called as far as I read the code.
> Likewise for poweroff.
>
Earlier, gxfb_suspend() performed each operation just for suspend event.
And as it was legacy code, it was invoked by pci_legacy_suspend() for
pci_pm_suspend(), pci_pm_freeze() and pci_pm_poweroff().
Thus, the code was wrapped inside "if" container:
if (state.event == PM_EVENT_SUSPEND) { }
After binding it with dev_pm_ops variable, pm->suspend() is invoked by just
pci_pm_suspend() which is required.
So I removed the "if" container and bind the callback with pm->suspend pointer
only.
Using SET_SYSTEM_PM_OPS will bring back the extra step of invoking gxfb_suspend()
for freeze and poweroff, even though the function will do nothing in that case.
Vaibhav Gupta
> Sam
>
> > +
> > static struct pci_driver gxfb_driver = {
> > .name = "gxfb",
> > .id_table = gxfb_id_table,
> > .probe = gxfb_probe,
> > .remove = gxfb_remove,
> > -#ifdef CONFIG_PM
> > - .suspend = gxfb_suspend,
> > - .resume = gxfb_resume,
> > -#endif
> > + .driver.pm = &gxfb_pm_ops,
> > };
> >
> > #ifndef MODULE
> > diff --git a/drivers/video/fbdev/geode/suspend_gx.c b/drivers/video/fbdev/geode/suspend_gx.c
> > index 1110a527c35c..8c49d4e98772 100644
> > --- a/drivers/video/fbdev/geode/suspend_gx.c
> > +++ b/drivers/video/fbdev/geode/suspend_gx.c
> > @@ -11,8 +11,6 @@
> >
> > #include "gxfb.h"
> >
> > -#ifdef CONFIG_PM
> > -
> > static void gx_save_regs(struct gxfb_par *par)
> > {
> > int i;
> > @@ -259,5 +257,3 @@ int gx_powerup(struct fb_info *info)
> > par->powered_down = 0;
> > return 0;
> > }
> > -
> > -#endif
> > --
> > 2.27.0
> >
> > _______________________________________________
> > dri-devel mailing list
> > dri-devel@lists.freedesktop.org
> > https://lists.freedesktop.org/mailman/listinfo/dri-devel
_______________________________________________
Linux-kernel-mentees mailing list
Linux-kernel-mentees@lists.linuxfoundation.org
https://lists.linuxfoundation.org/mailman/listinfo/linux-kernel-mentees
next prev parent reply other threads:[~2020-08-10 9:41 UTC|newest]
Thread overview: 38+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-08-05 18:07 [Linux-kernel-mentees] [PATCH v1 00/12] video: fbdev: use generic power management Vaibhav Gupta
2020-08-05 18:07 ` [Linux-kernel-mentees] [PATCH v1 01/12] fbdev: gxfb: " Vaibhav Gupta
2020-08-05 20:19 ` Bjorn Helgaas
2020-08-06 5:59 ` Vaibhav Gupta
2020-08-08 11:17 ` Sam Ravnborg
2020-08-10 9:39 ` Vaibhav Gupta [this message]
2020-08-10 9:44 ` Vaibhav Gupta
2020-08-10 16:54 ` Sam Ravnborg
2020-08-10 17:58 ` Vaibhav Gupta
2020-08-10 18:57 ` [Linux-kernel-mentees] [PATCH v2 00/12] video: fbdev: " Vaibhav Gupta
2020-08-10 18:57 ` [Linux-kernel-mentees] [PATCH v2 01/12] fbdev: gxfb: " Vaibhav Gupta
2020-08-16 20:16 ` Sam Ravnborg
2020-08-17 7:45 ` Vaibhav Gupta
2020-08-10 18:57 ` [Linux-kernel-mentees] [PATCH v2 02/12] fbdev: lxfb: " Vaibhav Gupta
2020-08-10 18:57 ` [Linux-kernel-mentees] [PATCH v2 03/12] fbdev: via-core: " Vaibhav Gupta
2020-08-10 18:57 ` [Linux-kernel-mentees] [PATCH v2 04/12] fbdev: aty: " Vaibhav Gupta
2020-08-10 18:57 ` [Linux-kernel-mentees] [PATCH v2 05/12] fbdev: aty128fb: " Vaibhav Gupta
2020-08-10 18:57 ` [Linux-kernel-mentees] [PATCH v2 06/12] fbdev: nvidia: " Vaibhav Gupta
2020-08-10 18:57 ` [Linux-kernel-mentees] [PATCH v2 07/12] fbdev: savagefb: " Vaibhav Gupta
2020-08-10 18:57 ` [Linux-kernel-mentees] [PATCH v2 08/12] fbdev: cyber2000fb: " Vaibhav Gupta
2020-08-10 18:57 ` [Linux-kernel-mentees] [PATCH v2 09/12] fbdev: i740fb: " Vaibhav Gupta
2020-08-16 20:24 ` Sam Ravnborg
2020-08-17 7:46 ` Vaibhav Gupta
2020-08-10 18:57 ` [Linux-kernel-mentees] [PATCH v2 10/12] fbdev: vt8623fb: " Vaibhav Gupta
2020-08-10 18:57 ` [Linux-kernel-mentees] [PATCH v2 11/12] fbdev: s3fb: " Vaibhav Gupta
2020-08-10 18:57 ` [Linux-kernel-mentees] [PATCH v2 12/12] fbdev: arkfb: " Vaibhav Gupta
2020-08-05 18:07 ` [Linux-kernel-mentees] [PATCH v1 02/12] fbdev: lxfb: " Vaibhav Gupta
2020-08-08 11:19 ` Sam Ravnborg
2020-08-05 18:07 ` [Linux-kernel-mentees] [PATCH v1 03/12] fbdev: via-core: " Vaibhav Gupta
2020-08-05 18:07 ` [Linux-kernel-mentees] [PATCH v1 04/12] fbdev: aty: " Vaibhav Gupta
2020-08-05 18:07 ` [Linux-kernel-mentees] [PATCH v1 05/12] fbdev: aty128fb: " Vaibhav Gupta
2020-08-05 18:07 ` [Linux-kernel-mentees] [PATCH v1 06/12] fbdev: nvidia: " Vaibhav Gupta
2020-08-05 18:07 ` [Linux-kernel-mentees] [PATCH v1 07/12] fbdev: savagefb: " Vaibhav Gupta
2020-08-05 18:07 ` [Linux-kernel-mentees] [PATCH v1 08/12] fbdev: cyber2000fb: " Vaibhav Gupta
2020-08-05 18:07 ` [Linux-kernel-mentees] [PATCH v1 09/12] fbdev: i740fb: " Vaibhav Gupta
2020-08-05 18:07 ` [Linux-kernel-mentees] [PATCH v1 10/12] fbdev: vt8623fb: " Vaibhav Gupta
2020-08-05 18:07 ` [Linux-kernel-mentees] [PATCH v1 11/12] fbdev: s3fb: " Vaibhav Gupta
2020-08-05 18:07 ` [Linux-kernel-mentees] [PATCH v1 12/12] fbdev: arkfb: " Vaibhav Gupta
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=20200810093948.GB6615@gmail.com \
--to=vaibhavgupta40@gmail.com \
--cc=FlorianSchandinat@gmx.de \
--cc=adaplas@gmail.com \
--cc=b.zolnierkie@samsung.com \
--cc=bhelgaas@google.com \
--cc=dilinger@queued.net \
--cc=dri-devel@lists.freedesktop.org \
--cc=helgaas@kernel.org \
--cc=linux-arm-kernel@lists.infradead.org \
--cc=linux-fbdev@vger.kernel.org \
--cc=linux-geode@lists.infradead.org \
--cc=linux-kernel-mentees@lists.linuxfoundation.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux@armlinux.org.uk \
--cc=paulus@samba.org \
--cc=sam@ravnborg.org \
--cc=vaibhav.varodek@gmail.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).