vincent.brosseau at bruker.fr
Thu Dec 12 08:17:22 PST 2013
Your explanation is very clear. Thanks a lot ! :-) And thanks for the link too.
De : dbus-bounces at lists.freedesktop.org [mailto:dbus-bounces at lists.freedesktop.org] De la part de Simon McVittie
Envoyé : jeudi 12 décembre 2013 16:10
À : dbus at lists.freedesktop.org
Objet : Re: dbus_message_set/get_data usage
On 12/12/13 13:18, Brosseau, Vincent wrote:
> You say that char[m][n] is not the same type as char*[n]. Why ?
> Arrays are pointers ( = *), and addresses are pointers (& = *).
Arrays are similar to pointers, but not the same.
> As far as I know, in my case ¶m.value[index] is like char** (isn't
> it ?)
No; if it was, your code wouldn't be crashing.
> dbus_message_iter_append_basic() knows that it's a string thanks to
> its second argument "DBUS_TYPE_STRING", then it knows that there's
> something else after the "h" (up to \0).
Suppose the "h" in "hello" is at memory address 0x88888888 and your machine is 32-bit (sizeof(char *) == 4, or 8 hex digits).
dbus_message_iter_append_basic(., DBUS_TYPE_STRING, .) wants to receive the address of a piece of memory containing 0x88888888 as its third argument.
¶m.value[index] is the address of the array param.value[index], which is the address of the first element of the array, which is 0x88888888. dbus_message_iter_append_basic() reads the bytes from
0x88888888 to 0x8888888b, and they are ('h','e','l','l'), which is probably not a valid pointer, leading to a segfault when it dereferences that "pointer".
If you passed param.value[index] as a function argument, the numeric value pushed onto the stack would *also* be 0x88888888.
If you have a char ** temporary variable "tmp" (let's say its address is 0x44444444), and you say "tmp = ¶m.value[index]", that sets the bytes (0x44444444 ... 0x44444447) to 0x88888888; then calling dbus_message_iter_append_basic (., DBUS_TYPE_STRING, tmp) puts
0x44444444 on the stack. That's the address of a piece of memory containing 0x88888888, as desired.
 or maybe in a CPU register, but the principle's the same
dbus mailing list
dbus at lists.freedesktop.org
More information about the dbus