I don't mind this change, it gives more flexibility. However, could you please give more details about the hyperviser software and version as well as provide binary dmi dump.

Thanks,
Anton

Today I collected a snippet of memory from a virtual machine.  The snippet contained the DMI information so I figured I'd run dmidecode against it to see if the information was what I expected it to be.   However, dmidecode failed due to a bus error after printing the summary information (SMBIOS version, number of structs, etc).   This was because the structure table address in the snippet is a physical memory address (e.g. 0xeb01f).  So I added a way to specify the expected base address (e.g. 0xeb000) if the binary input file.

Patch attached for consideration.

Note that I made the display of "Table at 0x%08X." unconditional because the information that it displays is useful at guessing what value to supply to --mem-base.