Re: [PATCH 05/10] kbuild: sort the list of symbols exported by the kernel (__ksymtab)

[Alan Jenkins <alan-jenkins@tuffmail.co.uk>]

Andrew Morton wrote:
> On Tue,  3 Nov 2009 10:06:17 +0000
> Alan Jenkins <alan-jenkins@tuffmail.co.uk> wrote:
>
>   
>> modpost of vmlinux.o now extracts the ksymtab sections and outputs
>> sorted versions of them as .tmp_exports-asm.S.  These sorted sections
>> are linked into vmlinux and the original unsorted sections are
>> discarded.
>>
>> This will allow modules to be loaded faster, resolving symbols using
>> binary search, without any increase in the memory needed for the
>> symbol tables.
>>
>> This does not affect the building of modules, so hopefully it won't
>> affect compile times too much.
>>
>> Minimally tested on ARM under QEMU emulator.
>> Build tested on blackfin; output of "size -A" unchanged.
>>     
>
> I'm getting a segfault from write_exports().  
>
> (gdb) bt
> #0  0x0000003e5f075510 in strlen () from /lib64/libc.so.6
> #1  0x0000003e5f045cb8 in vfprintf () from /lib64/libc.so.6
> #2  0x0000003e5f06683a in vsnprintf () from /lib64/libc.so.6
> #3  0x0000000000401897 in buf_printf (buf=0x7fff5514f5e0, 
>     fmt=0x7fff5514f4e0 "0x%02x ") at scripts/mod/modpost.c:1692
> #4  0x00000000004042c8 in main (argc=1024, argv=0x7fff5514f5e0)
>     at scripts/mod/modpost.c:2063
> (gdb) f 4
> #4  0x00000000004042c8 in main (argc=1024, argv=0x7fff5514f5e0)
>     at scripts/mod/modpost.c:2063
> 2063                    buf_printf(&buf, "__EXPORT_%s_SYMBOL(%s,"
> (gdb) p sym
> $1 = (struct symbol *) 0x65c9f0
> (gdb) p *sym
> $2 = {next = 0x64c3a0, module = 0x610010, crc = 4077789248, crc_valid = 1, 
>   weak = 0, vmlinux = 0, kernel = 0, preloaded = 0, function = 1, 
>   export = export_unknown, name = 0x65ca10 "simple_prepare_write"}
> (gdb) p sym->export
> $3 = export_unknown
> (gdb) p/d sym->export
> $4 = 5
>
> but section_names[] (which could be static in write_exports() btw) has
> only five entries.
>   

Thanks.  I can't work out why this would happen.  Could you show the 
options modpost is being run with (make V=1 will do)?  Also confirm 
whether this is "MODPOST vmlinux.o" or "MODPOST 1001 modules".


 > main (argc=1024

It looks like this is the step "MODPOST 1001 modules".  But that 
shouldn't run write_exports().  We only run modpost with "-x" for the 
step "MODPOST vmlinux.o".  Also, the vmlinux-only modpost is run first, 
so I wonder why it didn't hit the same problem.

I don't know why sym->vmlinux==0 either; simple_prepare_write() should 
always be built into vmlinux.  Perhaps modpost is being run on libfs.o 
for some strange reason.

Alan