wluma: Crash with AMD GPU

Yes it seems like I am all out of bugs. Thanks a lot for your help @maximbaz , @cyrinux!

Perfect, at least we will not be walking in the dark, but we have precise description of what to fix, with helpful links 😁

Will share some hints in case you want to try to play around, if not - no big deal, we will get to this eventually 😉

It crashes on mapping the memory here:

https://github.com/maximbaz/wluma/blob/a57c4f7bb7997d6bc58efaf73a99f00d5fe24212/src/frame/processor/vulkan.rs#L66

It tried to map the memory to the following allocated buffer:

https://github.com/maximbaz/wluma/blob/a57c4f7bb7997d6bc58efaf73a99f00d5fe24212/src/frame/processor/vulkan.rs#L165

Which was allocated in some memory that fits these requirements:

https://github.com/maximbaz/wluma/blob/a57c4f7bb7997d6bc58efaf73a99f00d5fe24212/src/frame/processor/vulkan.rs#L159

Specifically, I’m nearly positive that the bug is here:

https://github.com/maximbaz/wluma/blob/a57c4f7bb7997d6bc58efaf73a99f00d5fe24212/src/frame/processor/vulkan.rs#L161

Instead of picking the index of the right memory, we just pass 0, which on our laptop is the only available memory, so the choice is always right 😁

The fix could be as simple as trying to hardcode there 1 and seeing if it works, but I dont know if indices are actually incremental - but for sure there is an API call where we can get the right index for the memory that reports VK_MEMORY_PROPERTY_HOST_VISIBLE_BIT (as per the error message), and then pass the right index there.

If you do look into it, please keep us posted how it goes!

If you want to make a head start, here’s what you can do:

* install Vulkan validation layers (are you on Arch? `extra/vulkan-validation-layers`)

* `export VK_INSTANCE_LAYERS=VK_LAYER_KHRONOS_validation`

* `cargo run`

This is an extremely helpful tool, it will log all wrongdoings related to Vulkan. On my machine it produces no issues, but I’m willing to bet that on your desktop we will get some nice errors to fix.

Yes, I am on Arch.

VUID-vkMapMemory-memory-00682(ERROR / SPEC): msgNum: -330527817 - Validation Error: [ VUID-vkMapMemory-memory-00682 ] Object 0: handle = 0xfa21a40000000003, type = VK_OBJECT_TYPE_DEVICE_MEMORY; | MessageID = 0xec4c8bb7 | Mapping Memory without VK_MEMORY_PROPERTY_HOST_VISIBLE_BIT set: VkDeviceMemory 0xfa21a40000000003[]. The Vulkan spec states: memory must have been created with a memory type that reports VK_MEMORY_PROPERTY_HOST_VISIBLE_BIT (https://www.khronos.org/registry/vulkan/specs/1.2-extensions/html/vkspec.html#VUID-vkMapMemory-memory-00682)
    Objects: 1
        [0] 0xfa21a40000000003, type: 8, name: NULL
thread 'predictor-LG HDR 4K' panicked at 'Unable to compute luma percent: ERROR_MEMORY_MAP_FAILED', src/frame/capturer/wlroots.rs:124:26

Yes 5 is enough, it only failed one or two times with 4.

With https://github.com/maximbaz/wluma/commit/40445411c04abd812eb7cbb231d256d59f4bf953 the missing serial number should be fixed, together with having more debug logging to identify why displays are not being matched. Untested by hopefully will work 😛

DDC/I2C errors like the above is something @cyrinux experienced as well (I dont use external monitors so we’ve been testing on his setup), we concluded that it’s just transient errors that we can’t do anything about - although for him the errors only happened if he was messing with screen using ddcutil cli in parallel. So for you the errors occur even just by running wluma and nothing else?

To clarify, as long as the errors are transient, the app should just work, we log and ignore those errors.

Let me summarize to see if I remember it all correctly, the state with the latest main code:

• On desktop with external GPU, only capturer=none now works
• On laptop, capturer=wlroots now works with both only internal screen and with external display plugged in.

Everything correct?

Made some progress: One problem is, that my display has no serial. Because of that serial is not Some but None and the iteration does not even reach the display I actually want to select here: https://github.com/maximbaz/wluma/blob/4538b692274d27780ce1ac832a6b2156cb066810/src/brightness/ddcutil.rs#L60 After fixing that (substituted serial_number with model_id) it can actually match my display (no more warning) but still fails at getting the intial brightness

env RUST_LOG=debug ./target/debug/wluma
[2022-01-16T23:07:41Z DEBUG wluma] Using config: Config { frame: Frame { capturer: None, processor: Vulkan }, als: Time { thresholds_string_key: {}, thresholds: {16: "normal", 13: "bright", 7: "dark", 0: "night", 18: "dark", 20: "night", 9: "dim", 11: "normal"} }, output_by_type: OutputByType { backlight: [], ddcutil: [] }, keyboard: None, output: [DdcUtil(DdcUtilOutput { name: "LG HDR 4K" })] }
Continue adjusting brightness and wluma will learn your preference over time.
thread 'predictor-LG HDR 4K' panicked at 'Did not receive initial brightness value in time', src/predictor/controller.rs:69:25
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrac

@maximbaz Sure I would love to help with debugging the vulkan crash.

I have applied https://github.com/maximbaz/wluma/pull/28 and here is the new output if that helps:

env RUST_LOG=debug ./target/debug/wluma
[2022-01-16T20:55:29Z DEBUG wluma] Using config: Config { frame: Frame { capturer: None, processor: Vulkan }, als: Time { thresholds_string_key: {}, thresholds: {9: "dim", 0: "night", 16: "normal", 7: "dark", 11: "normal", 18: "dark", 13: "bright", 20: "night"} }, output_by_type: OutputByType { backlight: [], ddcutil: [] }, keyboard: None, output: [DdcUtil(DdcUtilOutput { name: "LG HDR 4K" })] }
Continue adjusting brightness and wluma will learn your preference over time.
[2022-01-16T20:55:30Z DEBUG wluma::brightness::ddcutil] display found: DisplayInfo { backend: I2cDevice, id: "22794", manufacturer_id: Some("GSM"), model_id: Some(30471), version: Some((1, 4)), serial: Some(544286), manufacture_year: Some(31), manufacture_week: Some(2), model_name: Some("LG HDR 4K"), serial_number: None, edid_data: Some([0, 255, 255, 255, 255, 255, 255, 0, 30, 109, 7, 119, 30, 78, 8, 0, 2, 31, 1, 4, 181, 60, 34, 120, 158, 62, 49, 174, 80, 71, 172, 39, 12, 80, 84, 33, 8, 0, 113, 64, 129, 128, 129, 192, 169, 192, 209, 192, 129, 0, 1, 1, 1, 1, 77, 208, 0, 160, 240, 112, 62, 128, 48, 32, 101, 12, 88, 84, 33, 0, 0, 26, 40, 104, 0, 160, 240, 112, 62, 128, 8, 144, 101, 12, 88, 84, 33, 0, 0, 26, 0, 0, 0, 253, 0, 56, 61, 30, 135, 56, 0, 10, 32, 32, 32, 32, 32, 32, 0, 0, 0, 252, 0, 76, 71, 32, 72, 68, 82, 32, 52, 75, 10, 32, 32, 32, 1, 247, 2, 3, 28, 113, 68, 144, 4, 3, 1, 35, 9, 7, 7, 131, 1, 0, 0, 227, 5, 192, 0, 230, 6, 5, 1, 82, 72, 93, 2, 58, 128, 24, 113, 56, 45, 64, 88, 44, 69, 0, 88, 84, 33, 0, 0, 30, 86, 94, 0, 160, 160, 160, 41, 80, 48, 32, 53, 0, 88, 84, 33, 0, 0, 26, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 44]), mccs_version: None, mccs_database: Database { entries: {} } }
[2022-01-16T20:55:30Z DEBUG wluma::brightness::ddcutil] display found: DisplayInfo { backend: I2cDevice, id: "22791", manufacturer_id: Some("BNQ"), model_id: Some(32561), version: Some((1, 3)), serial: Some(21573), manufacture_year: Some(26), manufacture_week: Some(8), model_name: Some("BenQ XL2411Z"), serial_number: Some("D2G05085SL0"), edid_data: Some([0, 255, 255, 255, 255, 255, 255, 0, 9, 209, 49, 127, 69, 84, 0, 0, 8, 26, 1, 3, 128, 53, 30, 120, 46, 157, 225, 166, 84, 84, 159, 38, 13, 80, 84, 165, 107, 128, 209, 192, 69, 124, 97, 124, 129, 128, 129, 188, 149, 60, 49, 124, 1, 1, 2, 58, 128, 24, 113, 56, 45, 64, 88, 44, 69, 0, 19, 42, 33, 0, 0, 30, 0, 0, 0, 255, 0, 68, 50, 71, 48, 53, 48, 56, 53, 83, 76, 48, 10, 32, 0, 0, 0, 253, 0, 56, 144, 30, 160, 33, 0, 10, 32, 32, 32, 32, 32, 32, 0, 0, 0, 252, 0, 66, 101, 110, 81, 32, 88, 76, 50, 52, 49, 49, 90, 10, 1, 222, 2, 1, 4, 0, 254, 91, 128, 160, 112, 56, 53, 64, 48, 32, 53, 0, 19, 42, 33, 0, 0, 26, 134, 111, 128, 160, 112, 56, 64, 64, 48, 32, 53, 0, 19, 42, 33, 0, 0, 26, 252, 126, 128, 136, 112, 56, 18, 64, 24, 32, 53, 0, 19, 42, 33, 0, 0, 30, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 199]), mccs_version: None, mccs_database: Database { entries: {} } }
[2022-01-16T20:55:30Z WARN  wluma] Skipping output 'LG HDR 4K' as it might be disconnected: Unable to find display
thread 'predictor-LG HDR 4K' panicked at 'Did not receive initial brightness value in time', src/predictor/controller.rs:69:25
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace

Edit: This is with capturer none.

To be sure, can you rollback your config to the one attach in the beginning of the issue and give us the few first lines of RUST_LOG=debug wluma ?

[2022-01-16T20:50:42Z DEBUG wluma] Using config: Config { frame: Frame { capturer: Wlroots, processor: Vulkan }, als: Time { thresholds_string_key: {}, thresholds: {9: "dim", 13: "bright", 7: "dark", 0: "night", 20: "night", 11: "normal", 16: "normal", 18: "dark"} }, output_by_type: OutputByType { backlight: [], ddcutil: [] }, keyboard: None, output: [DdcUtil(DdcUtilOutput { name: "LG HDR 4K" })] }
Continue adjusting brightness and wluma will learn your preference over time.
[2022-01-16T20:50:42Z DEBUG wluma::frame::capturer::wlroots] Using output 'Goldstar Company Ltd LG HDR 4K 0x00004E1E (DP-3)' for config 'LG HDR 4K'
thread 'predictor-LG HDR 4K' panicked at 'Unable to compute luma percent: ERROR_MEMORY_MAP_FAILED', src/frame/capturer/wlroots.rs:124:26
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace