Bug 203285
| Summary: | AMI Get Hostname command via ssif is throwing unspecified error. | ||
|---|---|---|---|
| Product: | Drivers | Reporter: | Balaji R G (balajirg) |
| Component: | Other | Assignee: | drivers_other |
| Status: | NEW --- | ||
| Severity: | normal | CC: | arnd, balajirg, minyard |
| Priority: | P1 | ||
| Hardware: | ARM | ||
| OS: | Linux | ||
| Kernel Version: | 5.1-rc4 | Subsystem: | |
| Regression: | No | Bisected commit-id: | |
I noticed this bug in mainline kernel v5.1-rc4. https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/drivers/char/ipmi/ipmi_ssif.c?h=v5.1-rc4 } else if (blocknum != ssif_info->multi_pos) { /* * Out of sequence block, just abort. Block * numbers start at zero for the second block, * but multi_pos starts at one, so the +1. */ result = -EIO; } Here blocknum + 1, should be added. This would be an IPMI bug not I2C bug, moving to the right component. Hi all, The above mentioned bug is fixed in the kernel upstream at "https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=55be8658c7e2feb11a5b5b33ee031791dbd23a69". So, I think we can close this issue. Thanks, Balaji R G |
We do hostname issue on BMC 3.09 and SUSL 15 (4.12.14-266.g5cc7bcc-default). We still met Unable to send RAW command (channel=0x0 netfn=0x32 lun=0x0 cmd=0x6b rsp=0xff): Unspecified error. linux-pgt4:~ # ipmitool raw 0x32 0x6B 1 0 Unable to send RAW command (channel=0x0 netfn=0x32 lun=0x0 cmd=0x6b rsp=0xff): Unspecified error linux-pgt4:~ # ipmitool -vvvvv raw 0x32 0x6B 1 0 Using ipmi device 0 Set IPMB address to 0x20 OpenIPMI Request Message Header: netfn = 0x6 cmd = 0x1 Sending request 0x1 to System Interface Got message: type = 1 channel = 0xf msgid = 0 netfn = 0x7 cmd = 0x1 data_len = 16 data = 002001030902bfd1b300020200000000 Iana: 46033 Running Get PICMG Properties my_addr 0x20, transit 0, target 0 OpenIPMI Request Message Header: netfn = 0x2c cmd = 0x0 OpenIPMI Request Message Data (1 bytes) 00 Sending request 0x0 to System Interface Got message: type = 1 channel = 0xf msgid = 1 netfn = 0x2d cmd = 0x0 data_len = 1 data = c1 Error response 0xc1 from Get PICMG Properities Running Get VSO Capabilities my_addr 0x20, transit 0, target 0 OpenIPMI Request Message Header: netfn = 0x2c cmd = 0x0 OpenIPMI Request Message Data (1 bytes) 03 Sending request 0x0 to System Interface Got message: type = 1 channel = 0xf msgid = 2 netfn = 0x2d cmd = 0x0 data_len = 1 data = c1 Invalid completion code received: Invalid command Acquire IPMB address Discovered IPMB address 0x0 Interface address: my_addr 0x20 transit 0:0 target 0x20:0 ipmb_target 0 RAW REQ (channel=0x0 netfn=0x32 lun=0x0 cmd=0x6b data_len=2) RAW REQUEST (2 bytes) 01 00 OpenIPMI Request Message Header: netfn = 0x32 cmd = 0x6b OpenIPMI Request Message Data (2 bytes) 01 00 Sending request 0x6b to System Interface Got message: type = 1 channel = 0xf msgid = 3 netfn = 0x33 cmd = 0x6b data_len = 1 data = ff Unable to send RAW command (channel=0x0 netfn=0x32 lun=0x0 cmd=0x6b rsp=0xff): Unspecified error linux-pgt4:~ # *************************************************************************** I checked the SLE15 ipmi drivers and found a bug in case of multi-part read in “ipmi_ssif.c” source file. Bug: File: “ipmi_ssif.c” Function/API : static void msg_done_handler(struct ssif_info *ssif_info, int result, unsigned char *data, unsigned int len) In case of mutlipart read, they are using ssif_info->multi_pos flag to check with the block number. The block number starts from zero(0) for second block but multi_pos starts from 1, so +1 should be added to the block number. Check the below code. Also attached the “ipmi_ssif.c” with this changes. Bug: } else if (blocknum != ssif_info->multi_pos) { Fix: } else if (blocknum + 1 != ssif_info->multi_pos) { /* * Out of sequence block, just abort. Block * numbers start at zero for the second block, * but multi_pos starts at one, so the +1. */ result = -EIO; } else {