Re: [PATCH 3/3] media: ov02c10: Accept a 26 MHz external clock

[Tomas Moro]

Hi Felipe,

First of all, thank you for the good work on this. 
I tested this series (v1, all three patches) on a Samsung Galaxy Book3
Pro, whose board feeds the sensor 26 MHz. Without the series the sensor
does not probe:

  ov02c10 i2c-OVTI02C1:00: error -EINVAL: external clock 26000000 is not supported
  ov02c10 i2c-OVTI02C1:00: probe with driver ov02c10 failed with error -22

Hardware
  DMI: SAMSUNG ELECTRONICS CO., LTD. / 960XFG / NP960XFG-KC2IT
  BIOS P07RGU.330.240529.ZQ, i7-1360P, IPU6 [8086:a75d]
  sensor ACPI HID OVTI02C1, clock 26 MHz (clk_summary)

Software
  7.2.6 (linux-cachyos, clang + ThinLTO). ov02c10.c there is identical
  to v7.2.6 stable, the three patches applied. Only ov02c10
  was rebuilt (out of tree). In-tree intel_ipu6 + intel_ipu6_isys.
  libcamera 0.7.2, simple pipeline, GPU soft ISP.

Results
  probe            OK
  controls         link_frequency 541666667, pixel_rate 216666666,
                   2 lanes; get_selection bounds 1928x1092,
                   crop 1920x1080
  streaming        1920x1092 at 40.70 fps, steady, no drops over
                   several 400-frame runs. That is ~35 % above 30 fps,
                   which matches the link frequency. Raw SGRBG10 1928x1092 capture works too.
  stream start     one "csi2-0 error: Frame sync error" every time
  first buffer     the first dequeued buffer is always empty
  orientation      image is rotated 180 degrees. This model is not in
                   the ipu-bridge upside-down DMI list; strings above.
  s2idle           7 cycles (rtcwake -m freeze), idle and right after
                   streaming: all resumed, sensor still bound, capture
                   works.
  s2idle while     refused by isys, not a hang becuse the running stream is
  streaming        wedged afterwards, a new open works:

  intel_ipu6_isys.isys intel_ipu6.isys.40: PM: dpm_run_callback(): isys_suspend [intel_ipu6_isys] returns -16
  intel_ipu6_isys.isys intel_ipu6.isys.40: PM: failed to suspend: error -16
  PM: Some devices failed to suspend, or early wake event detected
  PM: suspend exit
  intel_ipu6_isys.isys intel_ipu6.isys.40: stream stop time out
  intel_ipu6_isys.isys intel_ipu6.isys.40: stream close time out

Same tests with out-of-tree intel-ipu6-psys loaded and bound
(intel/ipu6-drivers 71bddb5, built unmodified against this kernel):
identical results, 4 more s2idle cycles all resumed. The s2idle hang from
intel/ipu6-drivers#381 does not reproduce here on 7.2.6.

Footnote, unrelated to this series: I saw one hard lockup (no trace) on
the first stream after rmmod of intel-ipu6-psys. I think that taking it out 
might have given me a corrupt state, had to reboot. 

Happy to test v2.

Tested-by: Mars-Wave <tmorolias@xxxxxxxxx> # Samsung Galaxy Book3 Pro NP960XFG (github.com/MarsWave)