Mesa 7.11 Release Candidate 1 Arrives

[bwat47]

Does this have any good changes for intel graphics? (i'm running ironlake on arch linux, so I will probably get this not too long after it goes stable)

[ahlaht]

Thanks, but it was easier than I thought. I had to change CPU scaling governor from ondemand to performance and OpenArena just flies.

Thanks, that's very useful information. Just yesterday I noticed I may also have that problem. At least there is something shady about it. I didn't properly test it yet, because I also found some other bugs (sigh). I wonder if there is never end to these bugs...

[kraftman]

Right. With only 1 power state the file is disabled.

Thanks, but it was easier than I thought. I had to change CPU scaling governor from ondemand to performance and OpenArena just flies.

[69acid69]

can't compile r300g with latest LLVM!

Hi,

for a few weeks I can't compile r300g drivers, using latest LLVM 3.0 (2.9.99.svn20110711-32.1). At the end of compiling I've got error message like this:
r300_dri.so.tmp: undefined reference to `LLVMAddTypeName'

So, any patch or help!?

Thanks.

[ahlaht]

then it doesn't work.

Right. With only 1 power state the file is disabled. But you could still enable it if you wanted. Just move these 3 lines:

diff --git a/drivers/gpu/drm/radeon/radeon_pm.c b/drivers/gpu/drm/radeon/radeon_pm.c
index aaa19dc..d3ef691 100644
--- a/drivers/gpu/drm/radeon/radeon_pm.c
+++ b/drivers/gpu/drm/radeon/radeon_pm.c
@@ -608,6 +608,10 @@ int radeon_pm_init(struct radeon_device *rdev)

        INIT_DELAYED_WORK(&rdev->pm.dynpm_idle_work, radeon_dynpm_idle_work_handler);

+       if (radeon_debugfs_pm_init(rdev)) {
+               DRM_ERROR("Failed to register debugfs file for PM!\n");
+       }
+
        if (rdev->pm.num_power_states > 1) {
                /* where's the best place to put these? */
                ret = device_create_file(rdev->dev, &dev_attr_power_profile);
@@ -621,10 +625,6 @@ int radeon_pm_init(struct radeon_device *rdev)
                rdev->acpi_nb.notifier_call = radeon_acpi_event;
                register_acpi_notifier(&rdev->acpi_nb);
 #endif
-               if (radeon_debugfs_pm_init(rdev)) {
-                       DRM_ERROR("Failed to register debugfs file for PM!\n");
-               }
-
                DRM_INFO("radeon: power management initialized\n");
        }

If you try this be prepared for a crash once you read the file, but I suspect it has been disabled unnecessarily.

[kraftman]

It should be enough if your kernel has CONFIG_DEBUG_FS=y.

Yes, the option is enabled.

[ahlaht]

If it's enough to do: mount -t debugfs none /sys/kernel/debug

then it doesn't work. I suppose my graphic card doesn't support the feature.

It should be enough if your kernel has CONFIG_DEBUG_FS=y.

From what I remember from the code I think radeon_pm_info should be available even if power management is disabled. (But I might be wrong too.)

[kraftman]

Make sure you have debugfs mounted.

If it's enough to do: mount -t debugfs none /sys/kernel/debug

then it doesn't work. I suppose my graphic card doesn't support the feature.

[whizse]

Thanks, but it seems I don't have such option - there's no such file or directory

Make sure you have debugfs mounted.

[kraftman]

That's what I had been wondering with my evergreen card for months until I found out the clock speeds are not set correctly on my system.

If you haven't already make sure the clocks have been set to something reasonable:
[CODE]
cat /sys/kernel/debug/dri/0/radeon_pm_info

Thanks, but it seems I don't have such option - there's no such file or directory