Page 1 of 1
idt was loaded but isr fail when exception occur[solved]
Posted: Mon Nov 26, 2007 5:41 pm
by shellex
hi, i am new and i meet a strange problem.
i have loaded idt successfully, but when exception occur, cpu can not route interrupt service function.
Code: Select all
void pic_install(){
// ICW1
OUTB(0x11, 0x20);
OUTB(0x11, 0xA0);
// ICW2
OUTB(0x20, 0x21);
OUTB(0x28, 0xA1);
// ICW3
OUTB(0x04, 0x21);
OUTB(0x02, 0xA1);
// ICW4
OUTB(0x01, 0x21);
OUTB(0x01, 0xA1);
OUTB(0xFF, 0x21); // IRQ0-IRQ7
OUTB(0xFF, 0xA1); // IRQ8-IRQ15
}
void fill_idt(int idx, int offset, int dpl, int type, short sel) {
_IDT idt_entry = {
(sel << 16) | (offset & 0xffff),
(offset & 0xffff0000) | 0x8000 | (dpl << 13) | (type << 8 )
};
g_idt[idx] = idt_entry;
}
void idt_install(){
/*install defined exception */
int i = 0;
for ( ; i < 256; ++i){
fill_idt(i, (int)test, 0, INTR_GATE_TYPE, CODE_SEL); //just a test
}
CLI();
_IDT_descr idt_descr = {IDT_SIZE, (int)g_idt}, t;
idt_descr.len = IDT_SIZE;
idt_descr.addr = (int)g_idt;
asm __volatile__ ("lidt %0\n\t"::"m"(idt_descr));
asm __volatile__ ("sidt %0\n\t"::"m"(t));
video_display_format("dump idt, len:%X, addr:%X\n", t.len, t.addr);
asm __volatile__ ("lidt %0\n\t"::"m"(idt_descr));
}
void test(){
video_display_str("just a test:\n");
HALT();
}
and this is my init function:
Code: Select all
void k_init(){
video_display_str("[!]install idt\n");
idt_install();
video_display_str("[!]install pic\n");
pic_install();
video_display_str("[!]enable interrupt\n");
STI();
}
this is bochs output:
Code: Select all
....
00133138297i[FDD ] read() on floppy image returns 0
00133161789e[KBD ] unsupported io write to keyboard port 64, value = bf
00133220750e[CPU ] interrupt(): not code segment
00133220750e[CPU ] interrupt(): not code segment
00133220750i[CPU ] protected mode
00133220750i[CPU ] CS.d_b = 32 bit
00133220750i[CPU ] SS.d_b = 32 bit
00133220750i[CPU ] | EAX=00000004 EBX=00009000 ECX=0009ff3d EDX=00000000
00133220750i[CPU ] | ESP=0009ffd0 EBP=0009fff8 ESI=00019200 EDI=00009200
00133220750i[CPU ] | IOPL=0 id vip vif ac vm RF nt of df IF tf sf ZF af PF cf
00133220750i[CPU ] | SEG selector base limit G D
00133220750i[CPU ] | SEG sltr(index|ti|rpl) base limit G D
00133220750i[CPU ] | CS:0008( 0001| 0| 0) 00000000 000fffff 1 1
00133220750i[CPU ] | DS:0010( 0002| 0| 0) 00000000 000fffff 1 1
00133220750i[CPU ] | SS:0010( 0002| 0| 0) 00000000 000fffff 1 1
00133220750i[CPU ] | ES:0010( 0002| 0| 0) 00000000 000fffff 1 1
00133220750i[CPU ] | FS:0010( 0002| 0| 0) 00000000 000fffff 1 1
00133220750i[CPU ] | GS:0010( 0002| 0| 0) 00000000 000fffff 1 1
00133220750i[CPU ] | EIP=0000011a (0000011a)
00133220750i[CPU ] | CR0=0x00000011 CR1=0 CR2=0x00000000
00133220750i[CPU ] | CR3=0x00000000 CR4=0x00000000
00133220750i[CPU ] >> idiv eax, dword ptr ss:[ebp+0xfffffffc] : F77DFC
00133220750e[CPU ] exception(): 3rd (13) exception with no resolution, shutdown status is 00h, resetting
00133220750i[SYS ] bx_pc_system_c::Reset(SOFTWARE) called
00133220750e[CPU ] CPU_LOOP bx_guard.interrupt_requested=1
Next at t=133220750
(0) [0x0000011a] 0008:011a (unk. ctxt): idiv eax, dword ptr ss:[ebp+0xfffffffc] ; f77dfc
this is bochs display:
Code: Select all
[!]install idt
dump idt, len:0x800, addr:0x10A0
[!]instal pic
[!]install interrupt
Posted: Mon Nov 26, 2007 6:20 pm
by Dandee Yuyo
00133161789e[KBD ] unsupported io write to keyboard port 64, value = bf
Bochs is still complaing about how you try to enable A20:
Code: Select all
enable_a20:
inb $0x64, %al
testb $0x2, %al
jnz enable_a20
movb $0xbf, %al
outb %al, $0x64
He doesn't like that 0xbf at all.
00133220750e[CPU ] interrupt(): not code segment
00133220750e[CPU ] interrupt(): not code segment
Your IDT seems to be bad. How are your IDT_descr and IDT structures/types defined?
Posted: Tue Nov 27, 2007 1:45 am
by shellex
the code enable A20 comes from a book named "The Undocumented PC"",
and i define IDT etc. as follow:
struct _interrupt_descriptor_table {
long low;
long high;
}
typedef struct _interrupt_descriptor_table _IDT;
struct _interrupt_descriptor_table_descr {
short len;
long addr;
} __attribute__((packed));
Posted: Tue Nov 27, 2007 2:49 am
by shellex
Bochs shut up now, i replace the code enable A20.:
enable_a20:
inb $0x92, %al
orb $0x92, %al
outb %al, $0x92
however, i still can not route interrupt service function
Code: Select all
void k_init(){
video_display_str("[!]install idt\n");
idt_install();
video_display_str("[!]install pic\n");
pic_install();
video_display_str("[!]enable interrupt\n");
STI();
// test div zero exception
int a = 0;
video_display_format("test div zero: %d\n", 4 / a);
}
[/code]
Posted: Tue Nov 27, 2007 5:49 am
by JamesM
I'm confused - Dandee Yuyo has quoted code that is not in your post - did you edit it?
Don't edit posts and take out important information- it makes them USELESS to read later!
Posted: Tue Nov 27, 2007 6:33 am
by shellex
JamesM wrote:I'm confused - Dandee Yuyo has quoted code that is not in your post - did you edit it?
Don't edit posts and take out important information- it makes them USELESS to read later!
i am sorry that make you confused. but i didn't remove any importan code from original code and i didn't edit my original post...
i just try to describe my problem better in my last post...
i add some code to test the div zero exception...
Posted: Tue Nov 27, 2007 7:00 am
by Dandee Yuyo
@James: It's a quote from another post from shellex:
http://www.osdev.org/phpBB2/viewtopic.php?t=15439
@Shellex: Don't have much time right now to debug your code, but the structures are fine. Mainly I wanted to be sure your were not using a short base for the IDT as you were using for the GDT in your previous post.
the code enable A20 comes from a book named "The Undocumented PC"",
Maybe it works on real hardware but not on Bochs, I dunno. You can try this method from the book on real hardware by writing something at 1Mb, and checking that it is not rolling over and being written at 0000:0000
Posted: Tue Nov 27, 2007 7:31 am
by shellex
e...well, i enable A20 in another way now. and no error msg about it in the output os bochs.
and what do "short base " mean?
i am sorry that my english is poor.
Posted: Tue Nov 27, 2007 8:43 am
by Dandee Yuyo
By short base I mean a 16 bits value for the IDT base as you were using for the GDT base in your other post that XCHG noticed. It should be 32 bits, so according to your _interrupt_descriptor_table_descr structure is ok.
Are you sure all the shifts you do for filling the IDT in fill_idt() are ok? Don't have my code here @ work to compare... (nor time to do so)
It helped me a lot to code a dumpDescriptor function (together with dumpSelector, dumpGDT, dumpIDT, dumpTSS, dumpMem and so on) that saved me hours of debugging.
Posted: Tue Nov 27, 2007 9:32 am
by Dandee Yuyo
Use unsigned values in fill_idt and in your structures. The shifting might not be working as you expect due to the sign bit.
INTR_GATE_TYPE should be 14 (0x0e) for interrupt gates.
CODE_SEL ( a multiple of 8 ) must point to a valid code selector in your GDT.
Also, check this out:
http://www.osdever.net/bkerndev/Docs/idt.htm
http://en.wikipedia.org/wiki/Interrupt_descriptor_table
http://www.acm.uiuc.edu/sigops/roll_you ... 6/idt.html
Posted: Thu Nov 29, 2007 6:53 am
by shellex
ï¼ Dandee Yuyo
well, i checked my code. it seems everything are right...
Posted: Thu Nov 29, 2007 8:46 am
by shellex
I solved it! i did it ! just reset GDT~~~
Thx All, All people here~
@Dandee Yuyo, Thank YOU~
Posted: Sat Dec 01, 2007 6:45 am
by Dandee Yuyo
! just reset GDT
