SEO

I've got a code, that looks a little like this:

#define MAXNAME 20
#define MAXPACKET 20000

struct package_info
{
unsigned short transport_id;
char filename[MAXNAME];
time_t modified;
};

struct segment_header
{
unsigned short transport_id;
unsigned short size;
unsigned int segm_num;
};

struct segment
{
struct segment_header header;
unsigned char buffer[MAXPACKET - sizeof(struct segment_header)];
};

There is also a class:

class C {
public:
int do_sth(int sd)
{
segment segm; // HERE (explained later)
// do some stuff
};
static void* fun(void* arg)
{
C *pc = (C*)arg;
pc->do_sth(6);
};
run()
{
pthread_create(... fun, this);
}
};

And in main() there is code resembling this:

int main()
{
C c;
c.run();
// wait for c to stop
}

When line "pc->do_sth(6)" is executed, my thread is signaled with SIGBUS.
Removing line with comment "HERE" causes the program to run smoothly,
although it can't do anything useful without it.

I use:
# g++ -v
Reading specs from /usr/local/lib/gcc/hppa2.0w-hp-hpux11.11/3.4.1/specs
Configured with:
/scratch/zack/pkgbuild/3.3.1/hpux-11/gcc-3.4.1/configure --enable-languages=
c,c++ --enable-threads=posix --disable-nls --with-gnu-as --without-gnu-ld --
with-as=/usr/local/bin/as --with-ld=/usr/ccs/bin/ld --with-libiconv-prefix=/
usr/local --prefix=/usr/local
Thread model: posix
gcc version 3.4.1

and compile with:

g++ -Wall -pedantic -o mcast -g3 *.cpp -pthread

Under gdb I've got the following error:

Program received signal SIGBUS, Bus error.
0x00003440 in CDeplClient::segm_receive(int) (this=0x40005cd0, sd=0)
at deplclient.cpp:126
126 {
(gdb)

If I'll insert line "segment segm;" into any other method, this method
instantly causes SIGBUS.

Any ideas what I am doing wrong?

TIA,

JD

"jd" <jedrzej_dudkiewicz@poczta.interia.pl> writes:

> int do_sth(int sd)
> {
> segment segm; // HERE (explained later)

The sizeof(segm) is 20000, that's not so big as to cause the thread
to run out of stack, unless you specified extremely small default
stack at pthread_create() time.

> pthread_create(... fun, this);

Perhaps the problem hides in the '...' above.
Post a minimal compilable test case, and your error will be revealed.

Cheers,
--
In order to understand recursion you must first understand recursion.
Remove /-nsp/ for email.

> > int do_sth(int sd)
> > {
> > segment segm; // HERE (explained later)
>
> The sizeof(segm) is 20000, that's not so big as to cause the thread
> to run out of stack, unless you specified extremely small default
> stack at pthread_create() time.

Actualy this was enough to run out of stack - there was about 50kB of data
on the stack already which I missed, both when I was preparing sample code
and looking through original one while trying to found a bug. Well, gotta
go, got a lot of things to learn

Thanks for help!

JD

PS I was thinking that it can be the case (too much data on stack), but then
I thought, that it would rise SIGSEGV... Wrong

Possibly an alignment error.

Try unioning a double over the
segment header to force 8-byte
alignment, and see what happens.



dk



"jd" <jedrzej_dudkiewicz@poczta.interia.pl> wrote in message
news:ctqqci$u2n$1@news.supermedia.pl...
> I've got a code, that looks a little like this:
>
> #define MAXNAME 20
> #define MAXPACKET 20000
>
> struct package_info
> {
> unsigned short transport_id;
> char filename[MAXNAME];
> time_t modified;
> };
>
> struct segment_header
> {
> unsigned short transport_id;
> unsigned short size;
> unsigned int segm_num;
> };
>
> struct segment
> {
> struct segment_header header;
> unsigned char buffer[MAXPACKET - sizeof(struct segment_header)];
> };
>
> There is also a class:
>
> class C {
> public:
> int do_sth(int sd)
> {
> segment segm; // HERE (explained later)
> // do some stuff
> };
> static void* fun(void* arg)
> {
> C *pc = (C*)arg;
> pc->do_sth(6);
> };
> run()
> {
> pthread_create(... fun, this);
> }
> };
>
> And in main() there is code resembling this:
>
> int main()
> {
> C c;
> c.run();
> // wait for c to stop
> }
>
> When line "pc->do_sth(6)" is executed, my thread is signaled with SIGBUS.
> Removing line with comment "HERE" causes the program to run smoothly,
> although it can't do anything useful without it.
>
> I use:
> # g++ -v
> Reading specs from /usr/local/lib/gcc/hppa2.0w-hp-hpux11.11/3.4.1/specs
> Configured with:
>
/scratch/zack/pkgbuild/3.3.1/hpux-11/gcc-3.4.1/configure --enable-languages=
>
c,c++ --enable-threads=posix --disable-nls --with-gnu-as --without-gnu-ld --
>
with-as=/usr/local/bin/as --with-ld=/usr/ccs/bin/ld --with-libiconv-prefix=/
> usr/local --prefix=/usr/local
> Thread model: posix
> gcc version 3.4.1
>
> and compile with:
>
> g++ -Wall -pedantic -o mcast -g3 *.cpp -pthread
>
> Under gdb I've got the following error:
>
> Program received signal SIGBUS, Bus error.
> 0x00003440 in CDeplClient::segm_receive(int) (this=0x40005cd0, sd=0)
> at deplclient.cpp:126
> 126 {
> (gdb)
>
> If I'll insert line "segment segm;" into any other method, this method
> instantly causes SIGBUS.
>
> Any ideas what I am doing wrong?
>
> TIA,
>
> JD
>
>