>>>But during modposting stage, I suspect a CRC check is done, on top of
>>>just the "extern" call, but because sys_clock_gettime() is not
>>>declared as EXPORT_SYMBOL().....the symbol become not accessible from
>>>a module. but if u compile the codes together with the kernel ELF
>>>image, compilation and runtime should not have any error. Problem
>>>now is modposting check done for a module......correct?
In my opinion,i think when u compile a module ,only compile a C file
,not link.like you do gcc -c *.c and make a object file.when you
insmod the module ,this moment begins to link and insmod performs like
a linker to do some symbol link check CRC etc.
2009/2/26 Peter Teoh <htmldeveloper@xxxxxxxxx>:
> feel free to correct me.
>
> since SYSCALL_DEFINE2() is ultimately defined as (include/linux/syscalls.h):
>
> #define SYSCALL_DEFINEx(x, name, ...) \
> asmlinkage long sys##name(__SC_DECL##x(__VA_ARGS__)); \
> static inline long SYSC##name(__SC_DECL##x(__VA_ARGS__)); \
> asmlinkage long SyS##name(__SC_LONG##x(__VA_ARGS__)) \
> { \
> __SC_TEST##x(__VA_ARGS__); \
> return (long) SYSC##name(__SC_CAST##x(__VA_ARGS__)); \
> } \
> SYSCALL_ALIAS(sys##name, SyS##name); \
> static inline long SYSC##name(__SC_DECL##x(__VA_ARGS__))
>
> so your sys_clock_gettime() will ultimately be declared as asmlinkage
> long sys_clock_gettime(), where asmlinkage will be mapped to something
> (irrelevant here) not related to static, which means that the symbol
> can always be accessed globally - just by issuing extern from the
> demand side. Correct?
>
> the error "symbol not found" must have occurred before this header
> file is included. perhaps after u included this header, then extern,
> then the symbol become accessible? I attempted, results as follows.
>
> C-wise, syntactically / semantically correct - compilation no errors:
>
> make -C /lib/modules/2.6.29-rc5/build
> M=/sda1/download/linux-2.6/for-linus/drivers/mymodule modules
> make[1]: Entering directory `/sda1/download/linux-2.6/for-linus'
> CC [M] /sda1/download/linux-2.6/for-linus/drivers/mymodule/peter.o
> /sda1/download/linux-2.6/for-linus/drivers/mymodule/peter.c: In
> function ‘init_hello_4’:
> /sda1/download/linux-2.6/for-linus/drivers/mymodule/peter.c:20:
> warning: ‘which_clock’ is used uninitialized in this function
> /sda1/download/linux-2.6/for-linus/drivers/mymodule/peter.c:20:
> warning: ‘tp’ is used uninitialized in this function
> Building modules, stage 2.
> MODPOST 1 modules
> WARNING: "sys_clock_gettime"
> [/sda1/download/linux-2.6/for-linus/drivers/mymodule/peter.ko]
> undefined!
> CC /sda1/download/linux-2.6/for-linus/drivers/mymodule/peter.mod.o
> LD [M] /sda1/download/linux-2.6/for-linus/drivers/mymodule/peter.ko
> make[1]: Leaving directory `/sda1/download/linux-2.6/for-linus'
>
> But during modposting stage, I suspect a CRC check is done, on top of
> just the "extern" call, but because sys_clock_gettime() is not
> declared as EXPORT_SYMBOL().....the symbol become not accessible from
> a module. but if u compile the codes together with the kernel ELF
> image, compilation and runtime should not have any error. Problem
> now is modposting check done for a module......correct?
>
> On Wed, Feb 25, 2009 at 6:08 PM, Ole Loots <ole@xxxxxxxxxxxxx> wrote:
>> Hello,
>>
>> I want to call the kernel function sys_clock_gettime, but when I just try to
>> call it, the linker complains about the unknown symbol/function
>> sys_clock_gettime, so i have hard coded a function pointer in my module...:
>>
>> long (*p_sys_clock_gettime)(clockid_t, struct timespec *tp) = 0xc004373c;
>>
>> That works fine for my system, but I know thats not the way to do it. How is
>> it done properly?
>>
>> Regards,
>> Ole
>>
>>
>> --
>> To unsubscribe from this list: send an email with
>> "unsubscribe kernelnewbies" to ecartis@xxxxxxxxxxxx
>> Please read the FAQ at http://kernelnewbies.org/FAQ
>>
>>
>
>
>
> --
> Regards,
> Peter Teoh
>
> --
> To unsubscribe from this list: send an email with
> "unsubscribe kernelnewbies" to ecartis@xxxxxxxxxxxx
> Please read the FAQ at http://kernelnewbies.org/FAQ
>
>
--
To unsubscribe from this list: send an email with
"unsubscribe kernelnewbies" to ecartis@xxxxxxxxxxxx
Please read the FAQ at http://kernelnewbies.org/FAQ
