gdb disassemble: show function offsets in base 16

Question

When disassembling functions, gdb will display memory addresses in base 16, but offsets in base 10.

Example:

(gdb) disassemble unregister_sysctl_table
Dump of assembler code for function unregister_sysctl_table:
   0x00037080 <+0>: push   %ebp
   0x00037081 <+1>: mov    %esp,%ebp
   0x00037083 <+3>: sub    $0x14,%esp
   0x00037086 <+6>: mov    %ebx,-0xc(%ebp)
   0x00037089 <+9>: mov    %esi,-0x8(%ebp)
   0x0003708c <+12>:mov    %eax,%ebx
   0x0003708e <+14>:mov    %edi,-0x4(%ebp)

The function offsets are the <+N> next to the address, and as you can see they are in base 10.

When the Linux kernel crashes, it displays a backtrace using base 16:

 [    0.524380]  [<c10381d5>] unregister_sysctl_table+0x65/0x70

It's very annoying to have to convert backtrace addresses from base 16 to base 10 to be able to find the desired instruction.

Can gdb be told to display disassembly output with base 16 offsets?

Answer 1

GDB currently uses hard-coded '%d' for the offset.

It's very annoying to have to convert backtrace addresses ... to be able to find the desired instruction

You do realize that you can simply do

x/i 0xc10381d5       # the crashing instruction (if looking at the inner frame)
x/i 0xc10381d5-5     # the call (if looking at caller frame)
x/10i 0xc10381d5-20  # context around the desired location