Sun Fire E25K Server
 
Sun Fire 12K Server
 
Sun Fire 15K Server
 
Sun Fire E20K Server
 

PLA-Support>Sun Systems>SPARC>Enterprise>SN-SPARC: SF-Exxk
 
_Old GCS Categories>Sun Microsystems>Servers>High-End Servers
 

Applies to:

Symptoms

 The System Controller (SC) platform messages file reports an error such as:

Apr 26 16:30:38 2003 sc-name frad[446]: [9912 442542960204 ERR SeepromInfoPro.cc 482] Bad section header on HPCI at IO14, bad element:tag, expected value: 8, actual value: 0

Changes

Cause

These error messages are the result of a corrupt hardware SEEPROM.

The SEEPROM resident on a FRU is normally 8192 bytes or 8K bytes in size. It is usually divided into two sections:

The first section is readable and writeable, while the second section is read only.
The first section (0) is normally 6K bytes in size while the second section (1) 2K bytes in size.
Each section begins with a 6-byte Section Header followed by one or more header expansions.
The Header expansions describe the type, offset, and length of each Segment in that section. 
The Manufacturer record which esmd tried to read via hwad is in the "SD" Segment residing on the second section, Section 1.
See the example below for a good header. The contents of Section 1 of an I/O Assembly (incidentally, also IO14) are displayed (extracted via redx through the Global I2C bus directly into the SEEPROM at I2C device address 50).

redx> ubuf size 8192
redx> i2crm sc 62 x50 0 x0 8192
redx> uba x1800 2048
0x001800:  08000110 73015344 000041B6 1F4000C0    ....s.SD..A..@..
0x001810:  00000000 00000000 00000000 00000000
................
[..]
0x001F40:  F80010B7 3C5F98A3 41535359 2C454342    ....<_..ASSY,ECB
0x001F50:  2C484F54 2D535741 50205043 492C5354    ,HOT-SWAP PCI,ST
0x001F60:  41524341 54000000 00000000 00000000    ARCAT...........
0x001F70:  00000000 00000000 00000000 00000000    ................
0x001F80:  00000000 00000000 00000000 00000000    ................
0x001F90:  00000000 00000000 454E4449 434F5454    ........ENDICOTT
0x001FA0:  204E5920 55534100 00000000 00000000     NY USA.........
0x001FB0:  00000000 00000000 00000000 00000000    ................
0x001FC0:  00000000 00000000 00000000 00000000    ................
0x001FD0:  00000000 00000000 35303135 33393732    ........50153972
0x001FE0:  32343034 3200A430 38353148 50434900    24042..0851HPCI.
0x001FF0:  00000000 00000000 0000000C 45A39851
............E..Q
Look at the above output starting at address 0x001800.
The Section 1 header is:
08  0001 10 73 01
^^  ^^^^ ^^ ^^ ^^
1   2    3  4  5
1. Section headers always start with "08".
2. The version is always 0001.
3. The Section header length is 10.
4. "73" is the CRC8 checksum.
5. The number of segments in this Section 1 is "1" which is the SD segment that is visible immediately after the Section 1 Header.
Look again at the platform messages error:
Apr 26 16:30:38 2003 sc-name frad[446]: [9912 442542960204 ERR SeepromInfoPro.cc 482] Bad section header on HPCI at IO14, bad element: tag, expected value: 8, actual value: 0
The correct value of the section header is "8" as explained above, but hwad read a value of "0".
Therefore,  the Section Header 1 tag is incorrect.

This SC should also report:
Apr 26 16:30:38 2003 orlocn01-sc0 esmd[1123]: [1985 442543869024 ERR FruAccess.cc 185] Failed to read the manifacturer record of fru IO14: rc=-3
This means that esmd fails to read the record.

Solution

Enter the Main Content

The I/O Assembly itself is good except the DFRUID data.
Replace the I/O assembly with a new one.
Replacing components for this type of failure is the only way to resolve the errors, but it is not a replacement that must be made immediately.
The failure is within the SEEPROM region, which is not used for data transactions.

Replace when time allows.

Additional Information:
Other components may report the error as well, but the result is the same.
The only resolution is replacing the part. An example of a different error message you may see is:
May 14 18:47:35 2003 sc-name frad[429]: [9912 1585343662509747 ERR SeepromInfoPro.cc 482] Bad section header on ECACHE at SB8/P0/E1, bad element: tag, expected value: 8, actual value: 0
This is the same error with the SEEPROM on E$ dimm SB8/P0/E1. The FRU in this case would be SB8.

This document was based on a starcat-support alias question answered by Tong-Pheng KOH, of PTS Server APAC, and in reference to EMEA Radiance case ID 36881744.

seeprom, frad, esmd, hwad

Previously Published As 72632

This solution has no attachment