media: i2c: cvs: Get the wake IRQ without claiming the GPIO

Since 7.1.13-200 (CONFIG_VIDEO_INTEL_CVS=m plus the driver backport) the Dell XPS 14 DA14260 (Panther Lake) has no sound: intel_cvs requests its wake GPIO with devm_gpiod_get() on probe, and every CS35L57 amplifier then fails cs35l56_get_speaker_id() with -EBUSY, so no card is registered. 7.1.12 without the driver creates the card; 7.2.4 and 7.2.5 fail the same way (rhbz#2531233 attachments 2157376/2157377 and comment 5). The evidence that the wake line is the amplifiers' shared speaker-ID GpioIo is laid out in rhbz#2529031 comment 5 (vendor driver comparison, debugfs consumers); DSDT confirmation is pending.

The wake line is only used as an IRQ. The patch takes the IRQ from the GpioInt entry with acpi_dev_gpio_irq_get_by(), as the I2C core does for client->irq and as the vendor driver does, without requesting the GPIO.

Posted to linux-media on 2026-09-08 with Fixes: 8e2b43d2 and Cc: stable, no review yet: https://lore.kernel.org/linux-media/20260908105717.496232-1-junjie.cao@intel.com/

Clean apply on fedora-7.2; core.o builds with allmodconfig on top of it. Not verified on the affected machine. The three reporters (rhbz#2529031, #2531233, #2531966) have been asked to boot with intel_cvs blacklisted, which removes the new GPIO consumer.

Bugzilla: https://bugzilla.redhat.com/show_bug.cgi?id=2529031

Signed-off-by: Junjie Cao junjie.cao@intel.com