OSDev.org

Posted: **Thu Jan 05, 2006 4:16 am**

I am debugging my ISR.

I placed two nop and a hlt before iret.
I use 's' command to step over each instruction
It goes well before ISR does iret.

But when it comes to IRET,the debugger freezed.Just like this:

Code: Select all

(0) [0x00009f18] 0010:00009f18 (unk. ctxt): nop                       ; 90
<bochs:132> s
Next at t=-675184644
(0) [0x00009f19] 0010:00009f19 (unk. ctxt): hlt                       ; f4
<bochs:133> s
Next at t=-675184643
(0) [0x00009f1a] 0010:00009f1a (unk. ctxt): add esp, 0x8              ; 81c40800
0000
<bochs:134> s
_(this is the cursor)

I have to use ctrl+c to make my program continue.Again,it comes to the timer interrupt's beginning.A loop!

by the way,if I enable keyboard and disable timer,there is similar problem.

I use the way of keeping print 'S' to see if my program is not in ISR.So,I think,when ISR returned ,it should continue to print 'S'.But the problem is screen freezed.

I think the debugger should return to where it was before the interrupt occured after doing 'iret'.

Can anyone explain this problem?

Posted: **Thu Jan 05, 2006 4:50 am**

wait a minute, i'm having trouble getting what's going on here ...

you put a "hlt" opcode within your interrupt handler ?

i don't remember of having any trouble with "iret" ...

Code: Select all

<bochs:49> disas 0x11b7f1 0x11b7f6
0011b7f1: (                    ): pop es                    ; 07
0011b7f2: (                    ): pop ds                    ; 1f
0011b7f3: (                    ): popad                     ; 61
0011b7f4: (                    ): pop ebp                   ; 5d
0011b7f5: (                    ): iretd                     ; cf
<bochs:27> break 0x11b7f5
<bochs:28> c
(0) Breakpoint 5, 0x11b7f5 in ?? ()
Next at t=302280708
(0) [0x0011b7f5] 0008:0011b7f5 (unk. ctxt): iretd                     ; cf 
<bochs:45> print-stack
   00c73f9c [0108ff9c]  13bcc9
   00c73fa0 [0108ffa0]  0008
   00c73fa4 [0108ffa4]  0212
   ...
<bochs:46> s
Next at t=302418023
(0) [0x0013bcc9] 0008:0013bcc9 (unk. ctxt): mov dword ptr ss:[esp], 0x13f730 ; c7042430f71300

just a last-minute though: what version of bochs are you using ?

Posted: **Thu Jan 05, 2006 8:37 am**

Try:

hlt
nop
nop

The reason is that when bochsdbg stop at hlt, and you press ctrl+c it jumps to the second one in the debugger. I don't think it actually skips a instruction, but anyway it should work with the above.

Posted: **Thu Jan 05, 2006 8:40 am**

If bochs is actually freezing up, I do not know it has to be a bug. If you mean it just continues running it is because you need two nops after the hlt.

OSDev.org

Can bochs trace "IRET"?

Can bochs trace "IRET"?

Re:Can bochs trace "IRET"?

Re:Can bochs trace "IRET"?

Re:Can bochs trace "IRET"?