Problems while retrieving SMART information via C602 SAS controller since 6.6.51

[Christian Kühnke <christian@xxxxxxxxxx>]

Hi all,

i recently noticed that on my oldish Fujitsu Primergy Server with a C602 
SAS controller and SATA disks, I would get strange SMART results from 
smartctl:

=== START OF READ SMART DATA SECTION ===
SMART Status not supported: Incomplete response, ATA output registers 
missing
SMART overall-health self-assessment test result: PASSED
Warning: This result is based on an Attribute check.

instead of

=== START OF READ SMART DATA SECTION ===
SMART overall-health self-assessment test result: PASSED

The last known working kernel was 6.6.47, the first kernel where I noted 
the error messages was 6.6.62.

Additionally, when retrieving the SMART data, the kernel would log:

Nov 21 12:47:24 mischa kernel: sas: Enter sas_scsi_recover_host busy: 1 
failed: 1
Nov 21 12:47:24 mischa kernel: sas: ata3: end_device-0:2: cmd error handler
Nov 21 12:47:24 mischa kernel: sas: ata1: end_device-0:0: dev error handler
Nov 21 12:47:24 mischa kernel: sas: ata2: end_device-0:1: dev error handler
Nov 21 12:47:24 mischa kernel: sas: ata4: end_device-0:3: dev error handler
Nov 21 12:47:24 mischa kernel: sas: ata3: end_device-0:2: dev error handler
Nov 21 12:47:24 mischa kernel: sas: --- Exit sas_scsi_recover_host: 
busy: 0 failed: 1 tries: 1

Today, I was able to narrow the problem down to the following kernel 
patch, first introduced into 6.6.51:

From 872f86e1757bbb0a334ee739b824e47c448f5ebc Mon Sep 17 00:00:00 2001
From: Igor Pylypiv <ipylypiv@xxxxxxxxxx>
Date: Tue, 2 Jul 2024 02:47:35 +0000
Subject: ata: libata-scsi: Check ATA_QCFLAG_RTF_FILLED before using 
result_tf

[ Upstream commit 816be86c7993d3c5832c3017c0056297e86f978c ]

qc->result_tf contents are only valid when the ATA_QCFLAG_RTF_FILLED flag
is set. The ATA_QCFLAG_RTF_FILLED flag should be always set for commands
that failed or for commands that have the ATA_QCFLAG_RESULT_TF flag set.

Reviewed-by: Hannes Reinecke <hare@xxxxxxx>
Reviewed-by: Damien Le Moal <dlemoal@xxxxxxxxxx>
Reviewed-by: Niklas Cassel <cassel@xxxxxxxxxx>
Signed-off-by: Igor Pylypiv <ipylypiv@xxxxxxxxxx>
Link: https://lore.kernel.org/r/20240702024735.1152293-8-ipylypiv@xxxxxxxxxx
Signed-off-by: Niklas Cassel <cassel@xxxxxxxxxx>
Signed-off-by: Sasha Levin <sashal@xxxxxxxxxx>
---
 drivers/ata/libata-scsi.c | 21 +++++++++++++++++++++
 1 file changed, 21 insertions(+)

(limited to 'drivers/ata/libata-scsi.c')

diff --git a/drivers/ata/libata-scsi.c b/drivers/ata/libata-scsi.c
index 27e0c87236ac04..c91f8746289f4a 100644
--- a/drivers/ata/libata-scsi.c
+++ b/drivers/ata/libata-scsi.c
@@ -242,10 +242,17 @@ void ata_scsi_set_sense_information(struct 
ata_device *dev,
  */
 static void ata_scsi_set_passthru_sense_fields(struct ata_queued_cmd *qc)
 {
+    struct ata_device *dev = qc->dev;
     struct scsi_cmnd *cmd = qc->scsicmd;
     struct ata_taskfile *tf = &qc->result_tf;
     unsigned char *sb = cmd->sense_buffer;

+    if (!(qc->flags & ATA_QCFLAG_RTF_FILLED)) {
+        ata_dev_dbg(dev,
+                "missing result TF: can't set ATA PT sense fields\n");
+        return;
+    }
+
     if ((sb[0] & 0x7f) >= 0x72) {
         unsigned char *desc;
         u8 len;
@@ -924,10 +931,17 @@ static void ata_to_sense_error(unsigned id, u8 
drv_stat, u8 drv_err, u8 *sk,
  */
 static void ata_gen_passthru_sense(struct ata_queued_cmd *qc)
 {
+    struct ata_device *dev = qc->dev;
     struct scsi_cmnd *cmd = qc->scsicmd;
     struct ata_taskfile *tf = &qc->result_tf;
     u8 sense_key, asc, ascq;

+    if (!(qc->flags & ATA_QCFLAG_RTF_FILLED)) {
+        ata_dev_dbg(dev,
+                "missing result TF: can't generate ATA PT sense data\n");
+        return;
+    }
+
     /*
      * Use ata_to_sense_error() to map status register bits
      * onto sense key, asc & ascq.
@@ -979,6 +993,13 @@ static void ata_gen_ata_sense(struct ata_queued_cmd 
*qc)
         ata_scsi_set_sense(dev, cmd, NOT_READY, 0x04, 0x21);
         return;
     }
+
+    if (!(qc->flags & ATA_QCFLAG_RTF_FILLED)) {
+        ata_dev_dbg(dev,
+                "missing result TF: can't generate sense data\n");
+        return;
+    }
+
     /* Use ata_to_sense_error() to map status register bits
      * onto sense key, asc & ascq.
      */
--
cgit 1.2.3-korg

I reverted this patch on this system and the problem has gone away. I 
have not enough kernel debugging skills to dig deeper into it. Does 
anybody have ideas?

Thanks in advance
Christian