Thank you for your contribution! Sashiko AI review found 1 potential issue to consider: - [High] drm/edid: Signed-off-by trailer uses handle instead of full real name --
commit ccbdbf24ebcae6dbb6cf32b1e44947510278e9c5 Author: ElXreno <[email protected]> > drm/edid: Read the monitor range from the DisplayID Adaptive-Sync block > > A DisplayID 2.x Adaptive-Sync data block (tag 0x2b) lists the refresh > rate ranges a display supports with an adaptive vertical total. Some > panels advertise their variable refresh range only there, and some > keep a narrower one in the base block range limits. The BOE > NE180QA1-MM1 is of the second kind: the range limits descriptor says > 48-120 Hz, the Adaptive-Sync block has three descriptors, > > 27 00 30 ef 00 00 > 27 00 30 77 00 00 > 27 00 30 3b 00 00 > > for 48-240, 48-120 and 48-60 Hz. A driver that checks a mode against > monitor_range then refuses VRR for the 240 Hz timing of that panel. > > Use the block in both cases. Without a range from the base block take > the range from the descriptors. With one, keep its minimum and only > raise the maximum, and only from descriptors that reach down into it, > so that the result never goes below what the base block promises. > > monitor_range stays one range per connector. Only descriptors for the > native panel range that declare the adaptive vertical total operation > mode count: a descriptor for fixed average vertical total alone does > not describe a range to pick a vertical total from. Where several > count, the highest minimum and the highest maximum are used. Without > a base range every rate in the result is then within the descriptor > that reaches highest. With one, every rate is within the base range > or within that descriptor, which overlaps it. > > The layout is taken from edid-decode (parse_displayid_adaptive_sync() > in utils/edid-decode/parse-displayid-block.cpp of v4l-utils), not > from the DisplayID text: a descriptor is 6 bytes plus the count in > bits 6:4 of the block revision, byte 0 has the native panel range > flag in bit 0 and the operation mode in bits 3:2 (1 for fixed average > and adaptive vertical total), byte 2 is the minimum refresh rate, > bytes 3-4 hold the maximum refresh rate minus one in ten bits. The > block is read from any section of structure version 2.0 or later, > which is what the panel above declares. > > This reaches every driver that reads monitor_range. A display with > such a block and no base block range now has a range where it had > none, which is the point for i915 and xe (their other VRR conditions > still apply). amdgpu scans DisplayID tag 0x25 on its own when the > minimum or the maximum is zero, and with a range from the > Adaptive-Sync block that scan is no longer reached. When the display > has no AMD VSDB, amdgpu takes monitor_range as its refresh rate > range, so it follows the raised maximum, and a DP or eDP display that > has only the Adaptive-Sync block can now pass its FreeSync test (a > range wider than 10 Hz and a sink that ignores the MSA timing). With > a VSDB that advertises FreeSync the range still comes from there. > > Signed-off-by: ElXreno <[email protected]> [Severity: High] Could the Author header and Signed-off-by trailer use a full real name instead of the handle ElXreno? Documentation/process/submitting-patches.rst requires a real name for the Developer's Certificate of Origin and does not permit pseudonyms or handles. -- Sashiko AI review ยท https://sashiko.dev/#/patchset/[email protected]?part=1
