- Jun 14, 2022
-
-
Ville Syrjälä authored
The VBIOS/GOP may not program the FDI M/n vs. dotclock entirely consistently. Eg. on a SNB Thinkpad X220 LVDS I see dotclock of 69.286 MHz (the best the DPLL can do) vs. FDI M/N 69.3 MHz (matches what the EDID actually declares). Signed-off-by:
Ville Syrjälä <ville.syrjala@linux.intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20220503182242.18797-15-ville.syrjala@linux.intel.com Reviewed-by:
Jani Nikula <jani.nikula@intel.com>
-
Ville Syrjälä authored
Pull the various iCLKIP parameters into a struct. Later on we'll reuse this during the state computation to determine the exact dotclock the hardware will be generating for us. v2: Don't lose the phaseinc calculation v3: Drop the misplaced '#include <intel_pch_refclk.h>' from intel_crt.c (Jani) Signed-off-by:
-
Ville Syrjälä authored
Extract intel_crtc_dotclock() from ddi_dotclock_get(). We'll reuse this during state computation in order to determine the actual final dotclcok after the DPLL computation has been done (which may not give us the exact same port_clock that we fed in). v2: Add the prototype Signed-off-by:
-
- Jun 10, 2022
-
-
José Roberto de Souza authored
Commit 3cf05076 ("drm/i915/bios: Split VBT data into per-panel vs. global parts") cause PSR to be disabled when enable_psr has the default value and there is at least one DP port that do not supports PSR. That was happening because intel_psr_init() is called for every DP port and then enable_psr is globaly set to 0 based on the PSR support of the DP port. Here dropping the enable_psr overwritten and using the VBT PSR value when enable_psr is set as default. Fixes: 3cf05076 ("drm/i915/bios: Split VBT data into per-panel vs. global parts") Cc: Ville Syrjälä <ville.syrjala@linux.intel.com> Cc: Jani Nikula <jani.nikula@intel.com> Cc: Jouni Högander <jouni.hogander@intel.com> Cc: Mika Kahola <mika.kahola@intel.com> Signed-off-by:
José Roberto de Souza <jose.souza@intel.com> Reviewed-by:
Jouni Högander <jouni.hogander@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20220608203344.513082-1-jose.souza@intel.com
-
- Jun 08, 2022
-
-
Ville Syrjälä authored
The eDP BDB block has gained yet another max link rate field. Let's parse it and consult it during the source rate filtering. v2: *20 instead of *2 to get the correct units (Jani) Reviewed-by:
-
Ville Syrjälä authored
We're not parsing the 5.4 Gbps value for the old eDP fast link training link rate, nor are we parsing the new fast link training link rate field. Remedy both. Also we'll now use the actual link rate instead of the DPCD BW register value. Note that we're not even using this information for anything currently, so should perhaps just nuke it all unless someone is planning on implementing fast link training finally... v2: Stop using the DPCD BW values (Jani) *20 instead of *2 to get the rate in correct units (Jani) Cc: Jani Nikula <jani.nikula@intel.com> Signed-off-by: -
Ville Syrjälä authored
We'll need to know the VBT panel_type before we can determine the maximum link rate for eDP. To that end move intel_dp_set_source_rates() & co. to be called after the per-panel VBT parsing has been done. intel_dp_mst_encoder_init() depends on the source rates so we'll have to do it a bit later as well. v2: Fix the intel_dp_mst_encoder_init() oops Reviewed-by: Jani Nikula <jani.nikula@intel.com> #v1 Signed-off-by:
-
Jani Nikula authored
There's an early return for !engine->kernel_context. Signed-off-by:
Tvrtko Ursulin <tvrtko.ursulin@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20220516081015.1058987-1-jani.nikula@intel.com
-
Jani Nikula authored
Fix: drivers/gpu/drm/i915/pxp/intel_pxp_debugfs.c:61:6: warning: symbol 'intel_pxp_debugfs_register' was not declared. Should it be static? Sort and remove the redundant pxp prefixes from the includes while at it. Signed-off-by:
-
Luca Coelho authored
The intel_dp_dsc_get_output_bpp() function outputs two lines of unconditional logs, which was okay when it was called only once. But now, we also call this function from intel_dp_mode_valid(), which is in turn called for every mode we need to validate. This causes a lot of useless noise. Remove the unconditional prints to avoid spamming the logs. Also remove one more print that is not unconditional, but is related. Signed-off-by:
Luca Coelho <luciano.coelho@intel.com> Signed-off-by:
-
Jani Nikula authored
Remove some of the unused helpers from i915_utils.h. Signed-off-by:
-
Jani Nikula authored
No longer needed after panel data was moved. Signed-off-by:
-
Jani Nikula authored
Add a dedicated file for the local functions around struct tasklet_struct. Far from ideal, but better placed in a dedicated file than i915_gem.h. Signed-off-by:
-
- Jun 06, 2022
-
-
José Roberto de Souza authored
This workaround brings some regressions to DG2 and if really necessary for DG2 an alternative workaround will be implemented. BSpec: 54077 Signed-off-by:
Matt Roper <matthew.d.roper@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20220602201730.199418-1-jose.souza@intel.com
-
- Jun 03, 2022
-
-
Jani Nikula authored
Split out audio registers to a header of its own to reduce the size of i915_reg.h. TODO: Remove direct audio register access from intel_ddi.c. However, unification of audio get config is cumbersome due to the audio enable bit being in the DP or HDMI registers on older platforms. Signed-off-by:
-
Balasubramani Vivekanandan authored
Voltage swing table updated for eDP HBR3 Bspec: 49291 Signed-off-by:
Balasubramani Vivekanandan <balasubramani.vivekanandan@intel.com> Reviewed-by:
Matt Atwood <matthew.s.atwood@intel.com> Signed-off-by:
-
- Jun 01, 2022
-
-
Ville Syrjälä authored
Some machines declare DRRS type = seamless, DRRS = no, DMRRS = yes. I *think* DMRRS stands for "dynamcic media refresh rate", and I suspect the way it's meant to work is that it lets the driver switch refresh rates to match the frame rate for media playback. Obviously for us all that kind of policy stuff is entirely up to userspace, so the only thing we may do is make the extra refresh rate(s) available. So let's treat this case as just static DRRS for now. In the future We might want to differentiate the "seamless w/ downclocking" vs. "seamless w/o downclocking" cases so that we could do seamless refresh rate changes for systems that only claim to support DMRRS. Closes: https://gitlab.freedesktop.org/drm/intel/-/issues/125 Acked-by:
-
Ville Syrjälä authored
Add a bunch of new struff we're missing in various BDB blocks. TODO: Bunch of these might actually need to be taken into use... v2: s/lfp_features/lfp_power/features/ (Jani) Reviewed-by:
-
Ville Syrjälä authored
Fix various typos around "aggressiveness". Note that the VBT spec also sometimes missspells it as "agressiveness" so I guess that's where some of the typos came from. Reviewed-by:
-
Ville Syrjälä authored
It seem that when dealing with VRR capable eDP panels we need to accept fixed modes with variable vblank length (which is what VRR varies dynamically). Typically the preferred mode seems to be a non-VRR more (lowish dotclock/refresh rate + short vblank). We also have examples where it looks like even the hblank length is a bit different between the preferred mode vs. VRR mode(s). So let's just accept anything that has matching hdisp+vdisp+flags. v2: Document that is_alt_drrs_mode() is a subset of is_alt_vrr_mode() (Jani) Closes: https://gitlab.freedesktop.org/drm/intel/-/issues/125 Reviewed-by:
-
Ville Syrjälä authored
To help with debugging DRRS/VRR panel init let's dump out all the fixed modes we rejected for whatever reason. Reviewed-by:
-
Ville Syrjälä authored
VBT seems to have an extra flag for VRR vs. not. Let's consult that for eDP panels. Reviewed-by:
-
Jani Nikula authored
Avoid bringing the entire machine down even if there's a bug that shouldn't happen, but won't corrupt the system either. Log them loudly and limp on. Signed-off-by:
Andrzej Hajda <andrzej.hajda@intel.com> Link: https://patchwork.freedesktop.org/patch/msgid/20220531162527.1062319-1-jani.nikula@intel.com
-
Ville Syrjälä authored
No idea why the DG2 PLL DP link frequency calculation is allowing a non-exact match. That makes no sense so get rid of it. Signed-off-by:
-
Ville Syrjälä authored
Don't see a real reson not to check hw.active and hw.enable in intel_pipe_config_compare(). We do have some checks for them at a higher level, but I think better check them also in intel_pipe_config_compare() in case something else doesn't do a thorough enough job. Also shuffle the mst_master_transcoder check next to the cpu_transcoder check for a bit of consistency. Signed-off-by:
-
Ville Syrjälä authored
Use the "[CRTC:%d:%s]'/etc. format for some of the modeset debugs so we know more about what has happened during the modeset state computation. Also tweak the connector bpp debug message a bit to make it less confusing. Signed-off-by:
-
Ville Syrjälä authored
Rename some of the 'pipe_config's to the more modern 'crtc_state'. Signed-off-by:
-
Ville Syrjälä authored
Use the state+crtc calling convention for intel_modeset_pipe_config() and othere related functions. Many of these need the full atomic state anyway so passing it all the way through is just nicer than having to worry about whether it can actually be extracted from eg. the crtc state passed in. Signed-off-by:
-
Ville Syrjälä authored
Deduplicate the drm_rect comparisons. We also drop the redundant pch_pfit.enabled check since the pch_pfit.dst rectanble will be zeroed anyway when the pfit is not enabled. v2: Document why we drop the enabled check (Jani) Signed-off-by:
-
Ville Syrjälä authored
Deduplicate the crtc_ timings comparisons. Signed-off-by:
-
Ville Syrjälä authored
The debugs in lower level DPLL code don't really provide any useful extra information AFAICS. Better just streamline the code and just put the necessary debugs (to identify at which step the modeset failed) into the higher level code. In addition we'll get the full state dump as well, which should hopefully have enough information to figure out what went wrong. Reviewed-by:
-
Ville Syrjälä authored
Split the DPLL state computation into a separate function from the current .get_dplls() which currently serves a dual duty by also reserving the shared DPLLs. v2: s/false/-EINVAL/ (Jani) Cc: Jani Nikula <jani.nikula@intel.com> Signed-off-by:
-
Vivek Kasireddy authored
Commit 30e114ef ("drm/i915/tc: Check for DP-alt, legacy sinks before taking PHY ownership") defaults any disconnected Type-C ports to TBT-alt mode which presents a problem (which could most likely result in a system hang) when userspace forces a modeset on a Type-C port that is wired for legacy HDMI. The following warning is seen when Weston forces a modeset on a disconnected legacy Type-C port (HDMI) on a TGL based Gigabyte system: (https://www.gigabyte.com/Mini-PcBarebone/GB-BSi3-1115G4-rev-10#ov ) Missing case (clock == 173000) WARNING: CPU: 1 PID: 438 at drivers/gpu/drm/i915/display/intel_ddi.c:245 icl_ddi_tc_enable_clock.cold+0x16a/0x1cf [i915] CPU: 1 PID: 438 Comm: kworker/u8:3 Tainted: G U W E 5.18.0-rc5-drm-tip+ #20 Hardware name: GIGABYTE GB-BSi3-1115G4/GB-BSi3-1115G4, BIOS F9 10/16/2021 Workqueue: i915_modeset intel_atomic_commit_work [i915] RIP: 0010:icl_ddi_tc_enable_clock.cold+0x16a/0x1cf [i915] Code: 74 6c 7f 10 81 fd d0 78 02 00 74 6d 81 fd b0 1e 04 00 74 70 48 63 d5 48 c7 c6 c0 7b ab c0 48 c7 c7 20 75 ab c0 e8 b8 b5 c1 f0 <0f> 0b 45 31 ed e9 fb fe ff ff 49 63 d5 48 c7 c6 80 7b ab c0 48 c7 RSP: 0018:ffff8882522c78f0 EFLAGS: 00010282 RAX: 0000000000000000 RBX: 0000000000000003 RCX: 0000000000000000 RDX: 0000000000000027 RSI: 0000000000000004 RDI: ffffed104a458f10 RBP: 0000000000011558 R08: ffffffffb078de4e R09: ffff888269ca748b R10: ffffed104d394e91 R11: 0000000000000000 R12: ffff888255a318f8 R13: 0000000000000002 R14: ffff888255a30000 R15: ffff88823ef00348 FS: 0000000000000000(0000) GS:ffff888269c80000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 00007fd7afa42000 CR3: 0000000255c02004 CR4: 00000000007706e0 PKRU: 55555554 Call Trace: <TASK> intel_ddi_pre_enable.cold+0x96/0x5bf [i915] intel_encoders_pre_enable+0x10e/0x140 [i915] hsw_crtc_enable+0x207/0x99d [i915] ? ilk_crtc_enable.cold+0x2a/0x2a [i915] ? prepare_to_wait_exclusive+0x120/0x120 intel_enable_crtc+0x9a/0xf0 [i915] skl_commit_modeset_enables+0x466/0x820 [i915] ? intel_commit_modeset_enables+0xd0/0xd0 [i915] ? intel_mbus_dbox_update+0x1ed/0x250 [i915] intel_atomic_commit_tail+0xf2d/0x3040 [i915] _raw_spin_lock_irqsave+0x87/0xe0 _raw_read_unlock_irqrestore+0x40/0x40 __update_load_avg_cfs_rq+0x70/0x5c0 __i915_sw_fence_complete+0x85/0x3b0 [i915] ? intel_get_crtc_new_encoder+0x190/0x190 [i915] ? sysvec_irq_work+0x13/0x90 ? asm_sysvec_irq_work+0x12/0x20 ? _raw_spin_lock_irq+0x82/0xd0 ? read_word_at_a_time+0xe/0x20 ? process_one_work+0x393/0x690 process_one_work+0x393/0x690 worker_thread+0x2b7/0x620 ? process_one_work+0x690/0x690 kthread+0x15a/0x190 ? kthread_complete_and_exit+0x20/0x20 ret_from_fork+0x1f/0x30 Continuing with the modeset without setting the DDI clock results in more warnings and eventually a system hang. This does not seem to happen with disconnected legacy or DP-alt DP ports because the clock rate defaults to 162000 (which is a valid TBT clock) during the link training process. Therefore, to fix this issue, this patch avoids setting disconnected Type-C legacy ports to TBT-alt mode which prevents the selection of TBT PLL when a modeset is forced. v2: (Imre) - Retain the check for legacy hotplug live status to account for incorrect VBTs. Cc: Imre Deak <imre.deak@intel.com> Cc: José Roberto de Souza <jose.souza@intel.com> Signed-off-by:
Vivek Kasireddy <vivek.kasireddy@intel.com> Reviewed-by:
Imre Deak <imre.deak@intel.com> Signed-off-by:
-
- May 28, 2022
-
-
Ville Syrjälä authored
Apparently when the VBT panel_type==0xff we should trawl through the PNPID table and check for a match against the EDID. If a match is found the index gives us the panel_type. Tried to match the Windows behaviour here with first looking for an exact match, and if one isn't found we fall back to looking for a match w/o the mfg year/week. v2: Rebase due to vlv_dsi changes v3: Adjust to .get_panel_type() vfunc Closes: https://gitlab.freedesktop.org/drm/intel/-/issues/5545 Reviewed-by:
-
Ville Syrjälä authored
Move the panel specific VBT parsing to happen during the output probing stage. Needs to be done because the VBT parsing will need to look at the EDID to determine the correct panel_type on some machines. We split the parsed VBT data (i915->vbt) along the same boundary. For the moment we just hoist all the panel specific stuff into connector->panel.vbt since that seems like the most convenient place for eg. the backlight code. Note that we simply drop the drrs type check from intel_drrs_frontbuffer_update() since that operates on the whole device rather than a specific connector/encoder. But the check was just a micro optimization so removing it doesn't actually mattter for correctness. TODO: Lot's of cleanup to be done in the future. Eg. most of the DSI stuff could probably be eliminated entirely and just parsed on demand during DSI init. v2: Note the intel_drrs_frontbuffer_update() change Signed-off-by:
-
Ville Syrjälä authored
Parsing the panel specific data (anything that depends on panel_type) from VBT is currently happening too early. Split the whole thing into global vs. panel specific parts so that we can start doing the panel specific parsing at a later time. v2: Clarify that this is about panel_type (Jani) Split out the leak checks (Jani) Reviewed-by: -
Ville Syrjälä authored
We use the "driver features" block for two different kinds of data: global data, and per panel data. Split the function into two parts along that line so that we can start doing the parsing in two different locations. Reviewed-by:
-
Ville Syrjälä authored
Disable the delayed VDD off work during the eDP probe. If we never turn off the VDD then we can't violate the panel's power sequencing delays despite not having read them out yet from the VBT. This is mostly a belt+suspenders type of thing since the the timeout we'd use for the delayed work should be long enough that this won't normally happen. But I don't really like relying on timeouts for correctless so might as well make sure. Signed-off-by:
-
Ville Syrjälä authored
During the eDP probe we may not yet know the panel_type used to index the VBT panel tables. So the initial eDP probe will have to be done without that, and thus we won't yet have the PPS delays from the VBT. Once the VBT has been fully parse we should reinit the PPS delays to make sure it's fully accounted for. Signed-off-by:
-
Ville Syrjälä authored
Split the PPS init to something we do at the start of the eDP probe and a second part we do at the end. Signed-off-by:
-
