NAS-142826 / 27.0.0-BETA.1 / drivetemp: fix error handling and parsing in the SCSI LOG SENSE path

[eschultz]

Jira: https://ixsystems.atlassian.net/browse/NAS-142826

Split out of #347 at review request. One commit, drivers/hwmon/drivetemp.c only, no ABI change and no behaviour change on a drive whose LOG SENSE succeeds. The temp1_crit to temp1_max retier and the subpage 02h read stay on #347.

All in drivetemp_retrieve_temp_log(), the SCSI log-page path added in 06ff843:

• scsi_execute_cmd() returns the positive scmd->result on CHECK CONDITION. hwmon_attr_show() only rejects negative returns, so an uninitialised stack long reaches userspace as the temperature. Normalise to -EIO, as drivetemp_scsi_command() already does since upstream 82163d6 (which we also carry on 6.12 as 42268d8).
• param_len is a u8 holding buf[i + 3] + 4. A length byte of 0xFC wraps it to 0 and the loop never advances, hanging the sysfs read with st->lock held. Use unsigned int.
• The loop checks i + param_len <= page_len but reads the temperature at buf[i + 5], which a parameter shorter than six bytes does not cover. Skip such parameters.

Base truenas/linux-6.18 @ 580caa3. Applies to truenas/linux-6.12 unchanged with Fixes: 43ff910ef5cc.

Verification: drivetemp.o cross-compiles clean for x86_64 including W=1 (x86_64_defconfig + CONFIG_TRUENAS=y + CONFIG_SENSORS_DRIVETEMP=m). checkpatch.pl --strict reports no errors or warnings. Not run on hardware; nothing here alters the success path.

[ixhamza]

Would be good to compact the commit messages and squash into one commit as well.

[eschultz]

Squashed to one commit with a compact message, on the same 580caa3 you reviewed. Picked up all three inline points; the param_len wrap is a real hang and is called out in the message. Compiles clean at W=1, checkpatch --strict clean.

[bugclerk]

This PR has been merged and conversations have been locked.
If you would like to discuss more about this issue please use our forums or raise a Jira ticket.

[ixhamza]

backport