Date: Fri, 30 Jan 2015 21:59:13 -0600 From: Elizabeth Myers <elizabeth@interlinked.me> To: Adrian Chadd <adrian@freebsd.org> Cc: freebsd-current <freebsd-current@freebsd.org> Subject: Re: Questions on adding backlight support for the i915 driver Message-ID: <54CC5311.9070604@interlinked.me> In-Reply-To: <CAJ-VmokWhk4NEjkvzU-g4McA1d4=%2BBrFYgrzpLWUZ5yQawaKsQ@mail.gmail.com> References: <54C883E7.4000300@interlinked.me> <54CBA0A4.30708@FreeBSD.org> <54CC0999.90500@interlinked.me> <3595567.zVLW9tlg0N@ralph.baldwin.cx> <54CC2981.3040909@interlinked.me> <CAJ-VmokWhk4NEjkvzU-g4McA1d4=%2BBrFYgrzpLWUZ5yQawaKsQ@mail.gmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
[-- Attachment #1 --]
On 01/30/15 19:31, Adrian Chadd wrote:
> So, this is one of the discussions that popped up in the linux side of
> things, that we will end up eventually pulling into freebsd when the
> i915 code is updated.
>
> The raw value is (a) different per setup, and (b) may be inverted to
> work correctly.
>
> I also don't know if linear values along that spectrum map to linear
> brightness levels.
>
> So yeah, my suggestion we expose an ACPI/acpi_video like list of valid
> percentages, and let the user set them from 0-100%. That way the
> details of how the actual backlight values get derived are hidden from
> userland.
>
>
>
> -adrian
I have a sort of "rough draft" of this. I've tested all the percentages
(Ivy Bridge) and they do seem to correlate linearly (and to the
intel_backlight userland program used by a lot of people). I haven't
been able to test on any other hardware as I don't have it, and I don't
know what chipsets require the value to be inverted, so I've not
implemented that.
If anyone else would like to help test, it'd be nice.
--
Cheers,
Elizabeth
[-- Attachment #2 --]
Index: i915/i915_debug.c
===================================================================
--- i915/i915_debug.c (revision 277875)
+++ i915/i915_debug.c (working copy)
@@ -724,7 +724,7 @@
for (i = 0; i < DRM_ARRAY_SIZE(error->ring); i++) {
struct drm_i915_error_object *obj;
-
+
if ((obj = error->ring[i].batchbuffer)) {
sbuf_printf(m, "%s --- gtt_offset = 0x%08x\n",
dev_priv->rings[i].name,
@@ -1590,6 +1590,38 @@
return (0);
}
+// NB - only works when y != 0!
+#define U_CEIL_DIV(x, y) ((x) + (y) - 1) / (y)
+
+static int
+i915_backlight(SYSCTL_HANDLER_ARGS)
+{
+ struct drm_device *dev;
+ drm_i915_private_t *dev_priv;
+ int error;
+ u32 val, max;
+
+ dev = arg1;
+ dev_priv = dev->dev_private;
+ if(dev_priv == NULL)
+ return (EBUSY);
+ DRM_LOCK(dev);
+ val = intel_panel_get_backlight(dev);
+ max = intel_panel_get_max_backlight(dev);
+ DRM_UNLOCK(dev);
+
+ // XXX specific devices, works fine on ivy bridge though
+ val = (val * 100) / max;
+ error = sysctl_handle_int(oidp, &val, 0, req);
+ if (error || !req->newptr)
+ return (error);
+ val = U_CEIL_DIV(max * val, 100);
+ DRM_LOCK(dev);
+ intel_panel_set_backlight(dev, val);
+ DRM_UNLOCK(dev);
+ return (0);
+}
+
static struct i915_info_sysctl_list {
const char *name;
int (*ptr)(struct drm_device *dev, struct sbuf *m, void *data);
@@ -1753,6 +1785,11 @@
CTLFLAG_RW, &i915_intr_pf, 0, NULL);
if (oid == NULL)
return (ENOMEM);
+ oid = SYSCTL_ADD_PROC(ctx, SYSCTL_CHILDREN(top), OID_AUTO,
+ "i915_backlight", CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_MPSAFE, dev,
+ 0, i915_backlight, "I", NULL);
+ if (oid == NULL)
+ return (ENOMEM);
error = drm_add_busid_modesetting(dev, ctx, top);
if (error != 0)
@@ -1764,6 +1801,5 @@
void
i915_sysctl_cleanup(struct drm_device *dev)
{
-
free(dev->sysctl_private, DRM_MEM_DRIVER);
}
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?54CC5311.9070604>