Networking NVIDIA Firmware Tools (MFT) Documentation

resourceparse Utility

The resourceparse tool parses and prints data segments content. The parser's output is used by NVIDIA representatives for debugging and troubleshooting.

This tool parses the output of the "resourcedump" tool. There are several parsing methods (see --parser):

  • ADB:
    This parsing method receives an ADB file and parses each segment data according to the ADB layout of the node with the segment_id attribute corresponding to the segment.

  • Map:
    This parsing method assumes that the provided resource-segments represent a memory map by DWORD pairs of address-value, and outputs each pair in a new line (similar to mstdump).

  • Compact map (cm):
    This parsing method assumes that the provided resource-segments represent a flat stream of value DWORDs, and receives the matching addresses from an mstdump-style CSV file (see --csv-file). It outputs each address-value pair in a new line (similar to mstdump).

The parsed segments are presented by a formatter (see --formatter):

  • basic:
    Presents each decoded field as a "field = value" line, under a header line naming the segment. This is the default, and the only formatter supported by parsing methods other than ADB.

  • pcie-events:
    Presents a dump of the PCIE_EVENTS resource as a PCIe event log: a block per event, with its time, class and name, followed by its location and details.

  • pcie-events-json:
    Serializes the same PCIe event log as a JSON document: the log header, an object per event with its location and details, and the issues raised while collecting them.

The tool applicable inputs for parsing can be the resourcedump outputs (bin file or the “human readable” format”), or the devlink json format output.


To parse the segments data in the most efficient way, the user must use the most suitable adb file. For the adb file, please contact Support.

resourceparse Usage

resourceparse -d DUMP_FILE [-p RESOURCE_PARSER] [-f FORMATTER] [--input_byte_order {be,le}] [--version] [-o OUT] [-v] -a ADB_FILE [-r]

where

-d, --dump-file

Location of the dump file used for parsing

--input_byte_order

Byte order of the DWORDs in a binary dump file, 'be' or 'le', ignored for a textual one. Default: 'be'

-p, --parser    

RESOURCE_PARSER: Available options: ['adb', 'map', 'cm', 'raw', 'menu']. 
Default: 'adb'. see Parsing Methods. 

-f, --formatter

FORMATTER: Available options: ['basic', 'pcie-events', 'pcie-events-json']. Default: 'basic'. Each parsing method supports a subset, see Formatters

-a, --adb-file

Location of the ADB file

-c, --csv-file

Location of the mstdump-style CSV file describing the addresses, either a ".csv" or a compressed ".csv.xz" file (cm parsing method)

-h, --help

Shows this help message and exit

--version

Shows the tool's version and exit

-o, --out

Location of the output file

--out-dir

Location of the output directory, a separate file is created for each segment (basic formatter only)

-r, --raw

Prints the raw data in addition to the parsed data (basic formatter only)

--hide-segment-header

Hide segment header during printing (basic formatter only)

--device

The device the events were read from, for the event log header (pcie-events and pcie-events-json formatters)

-v

Verbosity notice

Examples:

  • How to run basic parsing:

    # resourceparse --dump-file notice.txt --adb-file fw-4119-rel-16_23_2008.adb
    
    Parse 4 segments:
    
    --------------------------------------------------------------------------------
    
                        Segment - segment_info (0xfffe)
    
    segment_header.segment_type = 0xfffe
    
    segment_header.length_dw = 0x4
    
    dump_version = 0x0
    
    hw_version = 0x0
    
    fw_version = 0x1063232c
    
    --------------------------------------------------------------------------------
    
                        Segment - segment_command (0xfffa)
    
    segment_header.segment_type = 0xfffa
    
    segment_header.length_dw = 0x5
    
    vhca_id = 0x0
    
    segment_called = 0x2000
    
    index1 = 0x21
    
    index2 = 0x0
    
    num_of_obj1 = 0x0
    
    num_of_obj2 = 0x0
    
    --------------------------------------------------------------------------------
    
                        Segment - segment_notice (0xfff9)
    
    segment_header.segment_type = 0xfff9
    
    segment_header.length_dw = 0xc
    
    syndrome_id = 0x211
    
    notice[0] = 0x2000
    
    notice[1] = 0x21
    
    notice[2] = 0x0
    
    notice[3] = 0x0
    
    notice[4] = 0x496e7661
    
    notice[5] = 0x6c696420
    
    notice[6] = 0x52657300
    
    notice[7] = 0x0
    
    notice msg =  !Invalid Res
    
    --------------------------------------------------------------------------------
    
                        Segment - segment_terminate (0xfffb)
    
    segment_header.segment_type = 0xfffb
    
    segment_header.length_dw = 0x1
    
    --------------------------------------------------------------------------------
    


  • How to run  parsing with ‘raw’ and ‘verbosity’ options: 

    # resourceparse --dump-file notice.txt --adb-file fw-4119-rel-16_23_2008.adb –raw -v
    
    Notice - adb fw version 16.23.2008 is used for parsing while dump fw version is 16.99.9004
    
    Parse 4 segments:
    
    --------------------------------------------------------------------------------
    
                        Segment - segment_info (0xfffe)
    
    segment_header.segment_type = 0xfffe
    
    segment_header.length_dw = 0x4
    
    dump_version = 0x0
    
    hw_version = 0x0
    
    fw_version = 0x1063232c
    
    RAW DATA:
    
    DWORD [0-3]    :0x0004FFFE 0x00000000 0x00000000 0x1063232C
    
    --------------------------------------------------------------------------------
    
                        Segment - segment_command (0xfffa)
    
    segment_header.segment_type = 0xfffa
    
    segment_header.length_dw = 0x5
    
    vhca_id = 0x0
    
    segment_called = 0x2000
    
    index1 = 0x21
    
    index2 = 0x0
    
    num_of_obj1 = 0x0
    
    num_of_obj2 = 0x0
    
    RAW DATA:
    
    DWORD [0-3]    :0x0005FFFA 0x20000000 0x00000021 0x00000000
    
    DWORD [4]      :0x00000000
    
    --------------------------------------------------------------------------------
    
                        Segment - segment_notice (0xfff9)
    
    segment_header.segment_type = 0xfff9
    
    segment_header.length_dw = 0xc
    
    syndrome_id = 0x211
    
    notice[0] = 0x2000
    
    notice[1] = 0x21
    
    notice[2] = 0x0
    
    notice[3] = 0x0
    
    notice[4] = 0x496e7661
    
    notice[5] = 0x6c696420
    
    notice[6] = 0x52657300
    
    notice[7] = 0x0
    
    RAW DATA:
    
    DWORD [0-3]    :0x000CFFF9 0x00000211 0x00000000 0x00000000
    
    DWORD [4-7]    :0x00002000 0x00000021 0x00000000 0x00000000
    
    DWORD [8-11]   :0x496E7661 0x6C696420 0x52657300 0x00000000
    
    notice msg =  !Invalid Res
    
    --------------------------------------------------------------------------------
    
                        Segment - segment_terminate (0xfffb)
    
    segment_header.segment_type = 0xfffb
    
    segment_header.length_dw = 0x1
    
    RAW DATA:
    
    DWORD [0]      :0x0001FFFB
    
    --------------------------------------------------------------------------------
    


  • How to run parsing and save it into a file:

    # resourceparse --dump-file notice.txt --adb-file fw-4119-rel-16_23_2008.adb –out out_flie.txt
    
    write to file:  out_flie.txt
    


  • How to present a PCIe event log:

    # resourceparse --dump-file pcie_events.dump --adb-file /usr/lib64/mft/python_tools/nvpcie/data/pcie_event.adb --formatter pcie-events --device /dev/mst/mt4131_pciconf0
    
    PCIe Event Log
    Device        : /dev/mst/mt4131_pciconf0
    Event view    : critical
    FW version    : 20.33.0830
    Events        : 8
    
    [2026-08-13 06:30:00.000000]  [NON-CRITICAL]  **Fw information**
      fw major 0x18 | minor 0x7 | sub minor 0x141 | hash 0xdeadbeef
    
    [2026-08-13 06:30:01.500000]  [CRITICAL]  **Link status change**
      BDF  18:1f.2
      link UP x16 Gen5 | dl active | flit enabled
      perst none | directed no
    …
    


  • How to parse a compact dump written in little-endian, with the addresses of an installed mstdump database:

    # resourceparse --dump-file compact_dump.bin --input_byte_order le --parser cm --csv-file /usr/share/mft/mstdump_dbs/Vera.csv.xz --out vera.mstdump
    
    write to file:  vera.mstdump
    


Last updated: