Anyone used tcc?

[LinuxMan]

Hello. I try to replace gcc with TinyCC, but when i try compile one of my file(which fine compiled in gcc) i get error:
libc.c:36: incompatible types for redefinition of 'puthex'
My puthex:
void puthex(unsigned char byte)
{
   unsigned char lb, rb;

   lb=byte >> 4;

   rb=byte & 0x0F;

   puthexd(lb);
   puthexd(rb);

}
Anyone know how to resolve this?

[nick8325]

How do you declare puthex (in an .h file)? Perhaps the .h file defines it differently from the .c file.

[Pype.Clicker]

haven't used TinyCC either, but yes, it seems reasonable to think that you'll have to declare

Code: Select all

void puthex(unsigned char byte);

first if you intent to use it as a recursive function. For some reason, the prototype of the definition [tt]void puthex(...) {...}[/tt] seems to be ignored as long as the block isn't completely processed.

[LinuxMan]

Thanks! Yes, TCC need declared functions, instead of gcc that AFAIK don't need this always

[Candy]

Thanks! Yes, TCC need declared functions, instead of gcc that AFAIK don't need this always

GCC usually isn't pedantic. Try gcc -std=c99 -pedantic -pedantic-errors and try again

[LinuxMan]

Thx candy, now i know. Btw, i still have some problems with kernel. Kernel compiles without error but when i try to load they with grub system is reloaded . Linker script i use from http://www.mega-tokyo.com/osfaq/HigherHalfBareBones. Objdump -h of kernel:

Code: Select all

kernel.bin:     file format elf32-i386
Sections:
Idx Name          Size      VMA       LMA       File off  Algn
  0 .text         00001879  c0100000  00100000  00001000  2**5
                  CONTENTS, ALLOC, LOAD, READONLY, CODE
  1 .data         00001488  c0102000  00102000  00003000  2**5
                  CONTENTS, ALLOC, LOAD, DATA
  2 .bss          00004840  c01034a0  001034a0  00004488  2**5
                  ALLOC

I see that Align of .text is 2**5 instead of 2**4 in GCC variant. Anyone known how to fix this?

[Pype.Clicker]

that sounds more like a Triple Fault than something due to misalignment or whatever (actually, GRUB should be happy with whatever alignment constraints you'd place on it).

I suggest you run that in Bochs with a breakpoint at 1MB and see what suddenly goes wrong with your boot sequence (i assume it worked fine when compiled with GCC, right?) ... maybe some structure padding that happens differently on the two platforms ?

[LinuxMan]

How set breakpoint in bochs?

[Pype.Clicker]

1. make sure you have the internal debugger enabled in your bochs (./configure --enable-debugger)
2. run that bochs (bxdebug.exe for wusers, iirc), it should show a black screen and an invite such as

Next at t=0
(0) [0xfffffff0] f000:fff0 (unk. ctxt): jmp far f000:e05b ; ea5be000f0
<bochs:1>

3. enter your breakpoint with e.g. "pb 0x7c00" or "pb 0x100000"
4. continue execution with "c"

(0) Breakpoint 1, 0x7c00 in ?? ()
Next at t=2491287
(0) [0x00007c00] 0000:7c00 (unk. ctxt): jmp .+0x7c4a ; eb48
<bochs:3>

now the BIOS initialization is completed, you can start debugging your bootloader with step-by-step execution "s" or "step-over" execution "n" (to avoid entering BIOS functions again).

Or you might prefer to let GRUB initialize and only debugging once your kernel is loaded, which "pb 0x100000" then "c" should do nicely.

(erhm. if your code starts at 0x1000000 and not at some offset in the image like Clicker, of course.)

Note that the "p" in "pb" is for "physical". There's also "b <seg> <offset>" that can set e.g. a break point at logical address 0x8:0xC00000DE ...

[Pype.Clicker]

and a last-minute-trick: using a second terminal showing "objdump -drS your_elf_kernel.o | less" will definitely help in understanding the output of nasm disassembler

[LinuxMan]

Pype.Clicker, Oh. I need to recompile my bochs

[LinuxMan]

Code: Select all

(0) [0xfffffff0] f000:fff0 (unk. ctxt): jmp far f000:e05b         ; ea5be000f0
<bochs:1> pb 0xC0100000
<bochs:2> c
00000003740i[BIOS ] $Revision: 1.166 $ $Date: 2006/08/11 17:34:12 $
00000317560i[KBD  ] reset-disable command received
00000318690i[PIDE ] new BM-DMA address: 0xc000
00007100210i[BIOS ] int13_harddisk: function 41, unmapped device for ELDL=80
00007104970i[BIOS ] int13_harddisk: function 08, unmapped device for ELDL=80
00007109615i[BIOS ] *** int 15h function AX=00C0, BX=0000 not yet supported!
00195304420i[CPU0 ] protected mode
00195304420i[CPU0 ] CS.d_b = 32 bit
00195304420i[CPU0 ] SS.d_b = 32 bit
00195304420i[CPU0 ] EFER   = 0x00000000
00195304420i[CPU0 ] | RAX=0000000000000010  RBX=000000000002c480
00195304420i[CPU0 ] | RCX=00000000c01034c1  RDX=00000000000000cf
00195304420i[CPU0 ] | RSP=00000000c0107cb0  RBP=00000000c0107cb4
00195304420i[CPU0 ] | RSI=00000000000530fe  RDI=0000000000000000
00195304420i[CPU0 ] |  R8=0000000000000000   R9=0000000000000000
00195304420i[CPU0 ] | R10=0000000000000000  R11=0000000000000000
00195304420i[CPU0 ] | R12=0000000000000000  R13=0000000000000000
00195304420i[CPU0 ] | R14=0000000000000000  R15=0000000000000000
00195304420i[CPU0 ] | IOPL=0 id vip vif ac vm RF nt of df if tf SF zf af PF cf
00195304420i[CPU0 ] | SEG selector     base    limit G D
00195304420i[CPU0 ] | SEG sltr(index|ti|rpl)     base    limit G D
00195304420i[CPU0 ] |  CS:0008( 0001| 0|  0) 00000000 000fffff 1 1
00195304420i[CPU0 ] |  DS:0010( 0002| 0|  0) 00000000 000fffff 1 1
00195304420i[CPU0 ] |  SS:0010( 0002| 0|  0) 00000000 000fffff 1 1
00195304420i[CPU0 ] |  ES:0010( 0002| 0|  0) 00000000 000fffff 1 1
00195304420i[CPU0 ] |  FS:0010( 0002| 0|  0) 00000000 000fffff 1 1
00195304420i[CPU0 ] |  GS:0010( 0002| 0|  0) 00000000 000fffff 1 1
00195304420i[CPU0 ] |  MSR_FS_BASE:0000000000000000
00195304420i[CPU0 ] |  MSR_GS_BASE:0000000000000000
00195304420i[CPU0 ] | RIP=00000000c010006b (00000000c010006b)
00195304420i[CPU0 ] | CR0=0x80000011 CR1=0x0 CR2=0x0000000000000040
00195304420i[CPU0 ] | CR3=0x00102000 CR4=0x00000010
00195304420i[CPU0 ] >> mov ds, ax : 8ED8
00195304420e[CPU0 ] exception(): 3rd (14) exception with no resolution, shutdown status is 00h, resetting
00195304420i[SYS  ] bx_pc_system_c::Reset(SOFTWARE) called
00195304420i[APIC0] local apic in CPU 0 initializing
00195304420i[CPU0 ] CPU[0] is the bootstrap processor
00195304420e[CPU0 ] CPU_LOOP bx_guard.interrupt_requested=1
Next at t=195304421
bx_dbg_read_linear: physical memory read error (phy=0xc010006b, lin=0x00000000c010006b

Becouse in my link.ld i have ". = 0xC0100000;" so i think that pb must be 0xC0100000.

[Habbit]

"pb x" means "breakpoint at PHYSICAL address x"

what you need to do is either "lb 0xC0100000" (break at LINEAR address) or "pb 0x00100000", because GRUB loads you at physical 1 MiB, what you can page to whatever you want (linear address), but the physical address is the same

[LinuxMan]

No matter! I can don't set any breakpoints - s still have
bx_dbg_read_linear: physical memory read error (phy=0xc010006b, lin=0x00000000c010006b

[Habbit]

Set a linear breakpoint at the boot location (7c00) and step through your boot code (or, if you use GRUB, set the linear BP to your ASM entry point)