qcom_fg: handle SRAM readiness across suspend/resume

[tomaszduda23]

The fuel gauge SRAM may become ready during the suspend-to-resume transition. In this case, the system can enter standby before SRAM is ready, and the mem-avail interrupt will never fire. As a result, the driver misses the readiness event and its state becomes inconsistent.

Ensure that the SRAM is ready before suspending to avoid this issue.

A theoretical race condition may still exist, depending on the suspend ordering of related drivers, but this change prevents the common case of missed readiness.

[barni2000]

Remove [PATCH] from the commit title and please add Signed-off-by:
git commit -s --amend

[minlexx]

If this driver was taken from 8996 kernel, maybe it should be sent to 8996 kernel repo, where we all can then sync it from again..?

[barni2000]

If this driver was taken from 8996 kernel, maybe it should be sent to 8996 kernel repo, where we all can then sync it from again..?

8996 was get it from sdm845 so it should be sent back there maybe.
I have updated the driver from the sdm845 repo https://github.com/msm8953-mainline/linux/commits/barni2000/6.16.3/upgrade_fg/drivers/power/supply/qcom_fg.c

[tomaszduda23]

there is another fix qcom_fg.PATCH.
It requires

+static int qcom_fg_suspend(struct device *dev)
+{
+       struct qcom_fg_chip *chip = dev_get_drvdata(dev);
+
+       mutex_lock(&chip->lock);
+       chip->suspended = true;
+       mutex_unlock(&chip->lock);
+       flush_delayed_work(&chip->sram_release_access_work);
+
+       return 0;
+}

updated PR