Skip to contentSkip to conversation

help with inline assembly code?

DISCUSSION Chris Friesen 2 participants 5 messages Apr 24, 2009

lore
Share & tools
lore
#0 lore raw

Hi,

I've got a function that is used to overwrite opcodes in order to create self-modifying code. It worked just fine with previous compilers, but with gcc 4.3 it seems like it sometimes (but not always) causes problems when inlined. If I force it to never be inlined, it works fine.

First, here's the code:

void alter_opcode(unsigned long *addr, unsigned long opcode)

{
	asm volatile(
                "stw    %1,0(%0)	\n\t"
                "dcbf   0,%0		\n\t"
                "sync			\n\t"
                "icbi   0,%0,		\n\t"
                "isync			\n\t"
                    :: "r" (addr), "r" (opcode): "memory");
}

The symptom of the problem is a segfault on the "stw" instruction. I've verified that the address it's trying to write to is the expected address, and that the opcode being written is the expected opcode.

I assume I've mixed up the registers or constraints or something...anyone want to take a crack at it?

Chris

#1 lore raw
Quoted 15 lines · Chris Friesen

I've got a function that is used to overwrite opcodes in order to create self-modifying code. It worked just fine with previous compilers, but with gcc 4.3 it seems like it sometimes (but not always) causes problems when inlined. If I force it to never be inlined, it works fine.

First, here's the code:

void alter_opcode(unsigned long *addr, unsigned long opcode)

{
    asm volatile(
                "stw    %1,0(%0)    \n\t"
                "dcbf   0,%0        \n\t"
                "sync            \n\t"
                "icbi   0,%0,        \n\t"
                "isync            \n\t"
                    :: "r" (addr), "r" (opcode): "memory");
}
Quoted 1 line

The symptom of the problem is a segfault on the "stw" instruction. I've verified that the address it's trying to write to is the expected address,

Verified by looking at the address in "addr", or by looking at the reported faulting address?

Quoted 3 lines

and that the opcode being written is the expected opcode.

I assume I've mixed up the registers or constraints or something...anyone want to take a crack at it?

Is the compiler assigning r0 to addr? That will be treated as a literal zero instead. Try changing "r" (addr) to "b" (addr), or use stwx.

-Scott

#2 lore raw
Quoted 16 lines · Scott Wood
Quoted 15 lines · Chris Friesen

I've got a function that is used to overwrite opcodes in order to create self-modifying code. It worked just fine with previous compilers, but with gcc 4.3 it seems like it sometimes (but not always) causes problems when inlined. If I force it to never be inlined, it works fine.

First, here's the code:

void alter_opcode(unsigned long *addr, unsigned long opcode)

{
    asm volatile(
                "stw    %1,0(%0)    \n\t"
                "dcbf   0,%0        \n\t"
                "sync            \n\t"
                "icbi   0,%0,        \n\t"
                "isync            \n\t"
                    :: "r" (addr), "r" (opcode): "memory");
}
Quoted 3 lines
Quoted 1 line

The symptom of the problem is a segfault on the "stw" instruction. I've verified that the address it's trying to write to is the expected address,

Verified by looking at the address in "addr", or by looking at the reported faulting address?

Verified by running it in userspace under gdb, then looking at the registers listed in the disassembly and comparing it to the process maps.

Quoted 5 lines
Quoted 3 lines

and that the opcode being written is the expected opcode.

I assume I've mixed up the registers or constraints or something...anyone want to take a crack at it?

Is the compiler assigning r0 to addr? That will be treated as a literal zero instead. Try changing "r" (addr) to "b" (addr), or use stwx.

Bingo! Is there a constraint to tell the compiler to not use r0 for addr?

Chris

#3 lore raw
Quoted 4 lines · Chris Friesen
Quoted 1 line · Scott Wood

Is the compiler assigning r0 to addr? That will be treated as a literal zero instead. Try changing "r" (addr) to "b" (addr), or use stwx.

Bingo! Is there a constraint to tell the compiler to not use r0 for addr?

Yes, "b".

-Scott

#4 lore raw
Quoted 6 lines · Scott Wood
Quoted 3 lines · Chris Friesen
Quoted 1 line · Scott Wood

Is the compiler assigning r0 to addr? That will be treated as a literal zero instead. Try changing "r" (addr) to "b" (addr), or use stwx.

Bingo! Is there a constraint to tell the compiler to not use r0 for addr?

Yes, "b".

Doh. Sorry, apparently I can't read today. Both of those fixes seem to work.

Much appreciated.

Chris