Chapter 18 Debugging at the Machine-Instruction Level
This chapter describes how to use event management and process control commands at the machine-instruction level, how to display the contents of memory at specified addresses, and how to display source lines along with their corresponding machine instructions. The next, step, stop and trace commands each support a machine-instruction level variant: nexti, stepi, stopi, and tracei. The regs command can be used to print out the contents of machine registers or the print command can be used to print out individual registers.
This chapter is organized into the following sections:
Examining the Contents of Memory
Using addresses and the examine or x command, you can examine the content of memory locations as well as print the assembly language instruction at each address. Using a command derived from adb(1), the assembly language debugger, you can query for:
-
The address, using the = (equal sign) character; or,
-
The contents stored at an address, using the / (slash) character.
You can print the assembly commands using the dis and listi commands.
Using the examine or x Command
Use the examine command, or its alias x, to display memory contents or addresses.
Use the following syntax to display the contents of memory starting at addr for count items in format fmt. The default addr is the next one after the last address previously displayed. The default count is 1. The default fmt is the same as was used in the previous examine command, or X if this is the first command given.
The syntax for the examine command is:
examine [addr] [/ [count] [fmt ]]
To display the contents of memory from addr1 through addr2 inclusive, in format fmt:
examine addr1, addr2 [/ [fmt]]
Display the address, instead of the contents of the address in the given format:
examine addr = [fmt]
To print the value stored at the next address after the one last displayed by examine:
examine +/ i
To print the value of an expression, enter the expression as an address:
examine addr=format examine addr=
Addresses
The addr is any expression resulting in or usable as an address. The addr may be replaced with a + (plus sign) which displays the contents of the next address in the default format.
Example addresses are:
|
0xff99 |
An absolute address |
|
main |
Address of a function |
|
main+20 |
Offset from a function address |
|
&errno |
Address of a variable |
|
str |
A pointer-value variable pointing to a string |
Symbolic addresses used to display memory are specified by preceding a name with an ampersand (&). Function names can be used without the ampersand; &main is equal to main. Registers are denoted by preceding a name with a dollar sign ($).
Formats
The fmt is the address display format in which dbx displays the results of a query. The output produced depends on the current displayfmt. To change the display format, supply a different fmt code.
Set the fmt specifier to tell dbx how to display information associated with the addresses specified.
The default format set at the start of each dbx session is X, which displays an address/value as a 32-bit word in hexadecimal. The following memory display formats are legal.
|
i |
Display as an assembly instruction |
|
d |
Display as 16 bits (2 bytes) in decimal |
|
D |
Display as 32 bits (4 bytes) in decimal |
|
o |
Display as 16 bits (2 bytes) in octal. |
|
O |
Display as 32 bits (4 bytes) in octal. |
|
x |
Display as 16 bits (2 bytes) in hexadecimal. |
|
X |
Display as 32 bits (4 bytes) in hexadecimal. (default format) |
|
b |
Display as a byte in octal |
|
c |
Display as a character |
|
w |
Display as a wide character. |
|
s |
Display as a string of characters terminated by a null byte. |
|
W |
Display as a wide character. |
|
f |
Display as a single-precision floating point number. |
|
F, g |
Display as a double-precision floating point number. |
|
E |
Display as an extended-precision floating point number. |
|
ld, lD |
Display 32 bits (4 bytes) in decimal (same as D) |
|
lo, lO |
Display 32 bits (4 bytes) in octal (same as O) |
|
lx, LX |
Display 32 bits (4 bytes) in hexadecimal (same as X) |
|
Ld, LD |
Display 64 bits (8 bytes) in decimal |
|
Lo, LO |
Display 64 bits (8 bytes) in octal |
|
Lx, LX |
Display 64 bits (8 bytes) in hexadecimal |
Count
The count is a repetition count in decimal. The increment size depends on the memory display format.
Examples
The following examples show how to use an address with count and fmt options to display five successive disassembled instructions starting from the current stopping point.
(dbx) stepi stopped in main at 0x108bc 0x000108bc: main+0x000c: st %l0, [%fp - 0x14] (dbx) x 0x108bc/5i 0x000108bc: main+0x000c: st %l0, [%fp - 0x14] 0x000108c0: main+0x0010: mov 0x1,%l0 0x000108c4: main+0x0014: or %l0,%g0, %o0 0x000108c8: main+0x0018: call 0x00020b90 [unresolved PLT 8: malloc] 0x000108cc: main+0x001c: nop
(dbx) x &main/5i 0x08048988: main : pushl %ebp 0x08048989: main+0x0001: movl %esp,%ebp 0x0804898b: main+0x0003: subl $0x28,%esp 0x0804898e: main+0x0006: movl 0x8048ac0,%eax 0x08048993: main+0x000b: movl %eax,-8(%ebp)
Using the dis Command
The dis command is equivalent to the examine command with i as the default display format.
Here is the syntax for the dis command:
dis [addr][/count ]
The dis command without arguments displays 10 instructions starting at the address +. With only a count, the dis command displays count instructions starting at the address +.
Using the listi Command
To display source lines along with their corresponding assembly instructions, use listi, which is equivalent to list --i. See the discussion of list --i in Chapter 3, Viewing and Visiting Code".
(dbx) listi 13, 14 13 i = atoi(argv[1]); 0x0001083c: main+0x0014: ld [%fp + 0x48], %l0 0x00010840: main+0x0018: add %l0, 0x4, %l0 0x00010844: main+0x001c: ld [%l0], %l0 0x00010848: main+0x0020: or %l0, %g0, %o0 0x0001084c: main+0x0024: call 0x000209e8 [unresolved PLT 7: atoi] 0x00010850: main+0x0028: nop 0x00010854: main+0x002c: or %o0, %g0, %l0 0x00010858: main+0x0030: st %l0, [%fp - 0x8] 14 j = foo(i); 0x0001085c: main+0x0034: ld [%fp - 0x8], %l0 0x00010860: main+0x0038: or %l0, %g0, %o0 0x00010864: main+0x003c: call foo 0x00010868: main+0x0040: nop 0x0001086c: main+0x0044: or %o0, %g0, %l0 0x00010870: main+0x0048: st %l0, [%fp - 0xc]
(dbx) listi 13, 14 13 i = atoi(argv[1]); 0x080488fd: main+0x000d: movl 12(%ebp),%eax 0x08048900: main+0x0010: movl 4(%eax),%eax 0x08048903: main+0x0013: pushl %eax 0x08048904: main+0x0014: call atoi <0x8048798> 0x08048909: main+0x0019: addl $4,%esp 0x0804890c: main+0x001c: movl %eax,-8(%ebp) 14 j = foo(i); 0x0804890f: main+0x001f: movl -8(%ebp),%eax 0x08048912: main+0x0022: pushl %eax 0x08048913: main+0x0023: call foo <0x80488c0> 0x08048918: main+0x0028: addl $4,%esp 0x0804891b: main+0x002b: movl %eax,-12(%ebp)
Stepping and Tracing at Machine-Instruction Level
Machine-instruction level commands behave the same as their source level counterparts except that they operate at the level of single instructions instead of source lines.
Single-Stepping the Machine-Instruction Level
To single-step from one machine-instruction to the next machine-instruction, use nexti or stepi.
nexti and stepi behave the same as their source-code level counterparts: nexti steps over functions, stepi steps into a function called from the next instruction (stopping at the first instruction in the called function). The command forms are also the same. See next and step for a description.
The output from nexti and stepi differs from the corresponding source level commands in two ways. First, the output includes the address of the instruction at which the program is stopped (instead of the source code line number); secondly, the default output contains the disassembled instruction.
(dbx) func hand::ungrasp (dbx) nexti ungrasp +0x18: call support (dbx)
Tracing at the Machine-Instruction Level
Tracing techniques at the machine instruction level work the same as at the source code level, except you use tracei. For tracei, dbx executes a single instruction only after each check of the address being executed or the value of the variable being traced. tracei produces automatic stepi-like behavior: the program advances one instruction at a time, stepping into function calls.
When you use tracei, it causes the program to stop momentarily after each instruction while dbx checks for the address execution or the value of the variable or expression being traced. Using tracei can slow execution considerably.
For more information on trace and its event specifications and modifiers, see Chapter 5, Setting Breakpoints and Traces."
Here is the general syntax for tracei:
tracei event-specification [modifier]
Commonly used forms of tracei are:
|
tracei step |
Trace each instruction |
|
tracei next |
Trace each instruction, but skip over calls |
|
tracei at address |
Trace the given code address |
(dbx) tracei next -in main (dbx) cont 0x00010814: main+0x0004: clr %l0 0x00010818: main+0x0008: st %l0, [%fp - 0x8] 0x0001081c: main+0x000c: call foo 0x00010820: main+0x0010: nop 0x00010824: main+0x0014: clr %l0 .... .... (dbx) (dbx) tracei step -in foo -if glob == 0 (dbx) cont 0x000107dc: foo+0x0004: mov 0x2, %l1 0x000107e0: foo+0x0008: sethi %hi(0x20800), %l0 0x000107e4: foo+0x000c: or %l0, 0x1f4, %l0 ! glob 0x000107e8: foo+0x0010: st %l1, [%l0] 0x000107ec: foo+0x0014: ba foo+0x1c .... ....
Setting Breakpoints at Machine-Instruction Level
To set a breakpoint at machine-instruction level, use stopi. The command stopi accepts any event specification, using the syntax:
stopi event-specification [modifier]
Commonly used forms of the stopi command are:
stopi [at address ] [-if cond] stopi in function [-if cond]
Setting a Breakpoint at an Address
To set a breakpoint at a specific address:
(dbx) stopi at address
(dbx) nexti stopped in hand::ungrasp at 0x12638 (dbx) stopi at &hand::ungrasp (3) stopi at &hand::ungrasp (dbx)
Using the adb Command
The adb command allows you to enter commands in an adb(1) syntax. You may also enter adb mode which interprets every command as adb syntax. Most adb commands are supported.
For more information on the adb command, see the dbx online help.
Using the regs Command
The regs command lets you print the value of all the registers.
Here is the syntax for the regs command:
regs [-f][-F]
--f includes floating point registers (single precision). --F includes floating point registers (double precision); this is a SPARC only option.
dbx[13] regs -F current thread: t@1 current frame: [1] g0-g3 0x00000000 0x0011d000 0x00000000 0x00000000 g4-g7 0x00000000 0x00000000 0x00000000 0x00020c38 o0-o3 0x00000003 0x00000014 0xef7562b4 0xeffff420 o4-o7 0xef752f80 0x00000003 0xeffff3d8 0x000109b8 l0-l3 0x00000014 0x0000000a 0x0000000a 0x00010a88 l4-l7 0xeffff438 0x00000001 0x00000007 0xef74df54 i0-i3 0x00000001 0xeffff4a4 0xeffff4ac 0x00020c00 i4-i7 0x00000001 0x00000000 0xeffff440 0x000108c4 y 0x00000000 psr 0x40400086 pc 0x000109c0:main+0x4 mov 0x5, %l0 npc 0x000109c4:main+0x8 st %l0, [%fp - 0x8] f0f1 +0.00000000000000e+00 f2f3 +0.00000000000000e+00 f4f5 +0.00000000000000e+00 f6f7 +0.00000000000000e+00 ...
Platform-specific Registers
The following tables list platform-specific register names for SPARC and Intel that can be used in expressions.
SPARC Register Information
The following register information is for SPARC systems.
|
Register |
Description |
|---|---|
|
$g0 through $g7 |
Global registers |
|
$o0 through $o7 |
"out" registers |
|
$l0 through $l7 |
"local" registers |
|
$i0 through $i7 |
"in" registers |
|
$fp |
Frame pointer, equivalent to register $i6 |
|
$sp |
Stack pointer, equivalent to register $o6 |
|
$y |
Y register |
|
$psr |
Processor state register |
|
$wim |
Window invalid mask register |
|
$tbr |
Trap base register |
|
$pc |
Program counter |
|
$npc |
Next program counter |
|
$f0 through $f31 |
FPU "f" registers |
|
$fsr |
FPU status register |
|
$fq |
FPU queue |
The $f0f1 $f2f3 ... $f30f31 pairs of floating point registers are treated as having C "double" type (normally $fN registers are treated as C "float" type). These pairs can also be referred to as $d0 ... $d30.
The following additional registers are available on SPARC V9 and V8+ hardware:
$g0g1 through $g6g7 $o0o1 through $o6o7 $xfsr $tstate $gsr $f32f33 $f34f35 through $f62f63 ($d32 ... $$d62)
See the SPARC Architecture Reference Manual and the Sun-4 Assembly Language Reference Manual for more information on SPARC registers and addressing.
Intel Register Information
The following register information is for Intel systems.
|
Register |
Description |
|---|---|
|
$gs |
Alternate data segment register |
|
$fs |
Alternate data segment register |
|
$es |
Alternate data segment register |
|
$ds |
Data segment register |
|
$edi |
Destination index register |
|
$esi |
Source index register |
|
$ebp |
Frame pointer |
|
$esp |
Stack pointer |
|
$ebx |
General register |
|
$edx |
General register |
|
$ecx |
General register |
|
$eax |
General register |
|
$trapno |
Exception vector number |
|
$err |
Error code for exception |
|
$eip |
Instruction pointer |
|
$cs |
Code segment register |
|
$eflags |
Flags |
|
$es |
Alternate data segment register |
|
$uesp |
User stack pointer |
|
$ss |
Stack segment register |
Commonly used registers are also aliased to their machine independent names:
|
$sp |
Stack pointer; equivalent of $uesp |
|
$pc |
Program counter; equivalent of $eip |
|
$fp |
Frame pointer; equivalent of $ebp |
Registers for the 80386 lower halves (16 bits) are:
|
$ax |
General register |
|
$cx |
General register |
|
$dx |
General register |
|
$bx |
General register |
|
$si |
Source index register |
|
$di |
Destination index register |
|
$ip |
Instruction pointer, lower 16 bits |
|
$flags |
Flags, lower 16 bits |
The first four 80386 16-bit registers can be split into 8-bit parts:
|
$al |
Lower (right) half of register |
$ax |
|
$ah |
Higher (left) half of register |
$ax |
|
$cl |
Lower (right) half of register |
$cx |
|
$ch |
Higher (left) half of register |
$cx |
|
$dl |
Lower (right) half of register |
$dx |
|
$dh |
Higher (left) half of register |
$dx |
|
$bl |
Lower (right) half of register |
$bx |
|
$bh |
Higher (left) half of register |
$bx |
|
$fctrl |
Control register |
|
$fstat |
Status register |
|
$ftag |
Tag register |
|
$fip |
Instruction pointer offset |
|
$fcs |
Code segment selector |
|
$fopoff |
Operand pointer offset |
|
$fopsel |
Operand pointer selector |
|
$st0 through $st7 |
Data registers |
- © 2010, Oracle Corporation and/or its affiliates