Removing padding from structure in kernel module - gcc

Question

I am compiling a kernel module, containing a structure of size 34, using the standard command.
make -C /lib/modules/$(KVERSION)/build M=$(PWD) modules
The sizeof(some_structure) is coming as 36 instead of 34 i.e. the compiler is padding the structure.
How do I remove this padding?
Running make V=1 shows the gcc compiler options passed as
make -I../inc -C /lib/modules/2.6.29.4-167.fc11.i686.PAE/build M=/home/vishal/20100426_eth_vishal/organised_eth/src modules
make[1]: Entering directory `/usr/src/kernels/2.6.29.4-167.fc11.i686.PAE'
test -e include/linux/autoconf.h -a -e include/config/auto.conf || ( \
echo; \
echo " ERROR: Kernel configuration is invalid."; \
echo " include/linux/autoconf.h or include/config/auto.conf are missing."; \
echo " Run 'make oldconfig && make prepare' on kernel src to fix it."; \
echo; \
/bin/false)
mkdir -p /home/vishal/20100426_eth_vishal/organised_eth/src/.tmp_versions ; rm -f /home/vishal/20100426_eth_vishal/organised_eth/src/.tmp_versions/*
make -f scripts/Makefile.build obj=/home/vishal/20100426_eth_vishal/organised_eth/src
gcc -Wp,-MD,/home/vishal/20100426_eth_vishal/organised_eth/src/.eth_main.o.d -nostdinc -isystem /usr/lib/gcc/i586-redhat-linux/4.4.0/include -Iinclude -I/usr/src/kernels/2.6.29.4-167.fc11.i686.PAE/arch/x86/include -include include/linux/autoconf.h -D__KERNEL__ -Wall -Wundef -Wstrict-prototypes -Wno-trigraphs -fno-strict-aliasing -fno-common -Werror-implicit-function-declaration -Os -m32 -msoft-float -mregparm=3 -freg-struct-return -mpreferred-stack-boundary=2 -march=i686 -mtune=generic -Wa,-mtune=generic32 -ffreestanding -DCONFIG_AS_CFI=1 -DCONFIG_AS_CFI_SIGNAL_FRAME=1 -pipe -Wno-sign-compare -fno-asynchronous-unwind-tables -mno-sse -mno-mmx -mno-sse2 -mno-3dnow -Iarch/x86/include/asm/mach-generic -Iarch/x86/include/asm/mach-default -Wframe-larger-than=1024 -fno-stack-protector -fno-omit-frame-pointer -fno-optimize-sibling-calls -g -pg -Wdeclaration-after-statement -Wno-pointer-sign -fwrapv -fno-dwarf2-cfi-asm -DTX_DESCRIPTOR_IN_SYSTEM_MEMORY -DRX_DESCRIPTOR_IN_SYSTEM_MEMORY -DTX_BUFFER_IN_SYSTEM_MEMORY -DRX_BUFFER_IN_SYSTEM_MEMORY -DALTERNATE_DESCRIPTORS -DEXT_8_BYTE_DESCRIPTOR -O0 -Wall -DT_ETH_1588_051 -DALTERNATE_DESCRIPTORS -DEXT_8_BYTE_DESCRIPTOR -DNETHERNET_INTERRUPTS -DETH_IEEE1588_TESTS -DSNAPTYPSEL_TMSTRENA_TEVENTENA_TESTS -DT_ETH_1588_140_147 -DLOW_DEBUG_PRINTS -DMEDIUM_DEBUG_PRINTS -DHIGH_DEBUG_PRINTS -DMODULE -D"KBUILD_STR(s)=#s" -D"KBUILD_BASENAME=KBUILD_STR(eth_main)" -D"KBUILD_MODNAME=KBUILD_STR(conxt_eth)" -c -o /home/vishal/20100426_eth_vishal/organised_eth/src/eth_main.o /home/vishal/20100426_eth_vishal/organised_eth/src/eth_main.c

Answer 1

If using GCC, you can use the packed attribute on your structure to prevent padding:
struct foo
{
void * bar;
}
__attribute__( ( packed ) );

Answer 2

#pragma pack might work

Answer 3

I suspect that GCC is forcing the total structure to be aligned onto a 32 bit boundary, so its size is a multiple of 4.
Imagine the following.
struct foo
{
void * bar ;
some other stuff .....
};
struct foo my_foo_array[10];
Then if the sizeof(struct foo) is not a multiple of 4 then.
my_foo_array[0].bar has a different memory alignment to my_foo_array[1].bar. The processor would need to perform 2 32 bit memory accesses in order to access all four bytes of my_foo_array[1].bar. x86 processors will do this reassembly of misaligned 32 bit values but most other processors will throw some form of bus error exception which is not good.
The packed attribute signals how the elements of the structure are packed with respect to each other, but in normal operation the start of structure needs to placed on a 32 bit aligned address.
I hope this explains things a little better.