| 1 |
On 1/3/07, Iain Buchanan <iaindb@××××××××××××.au> wrote: |
| 2 |
> Hi all, |
| 3 |
> |
| 4 |
> I was just trying to compile one of my programs that includes |
| 5 |
> linux/threads.h for the #define PID_MAX_DEFAULT, however, threads.h |
| 6 |
> isn't there anymore! |
| 7 |
|
| 8 |
Hmm, on linux the maximum PID can be changed dynamically at run time. |
| 9 |
Wouldn't it be better to change this to read /proc/sys/kernel/pid_max? |
| 10 |
|
| 11 |
> I've just upgraded linux-headers to 2.6.19 - is this the problem? |
| 12 |
|
| 13 |
It seems so. From kernel-2.eclass: |
| 14 |
|
| 15 |
# 2.6.18 introduces headers_install which means we dont need any |
| 16 |
# of this crap anymore :D |
| 17 |
if kernel_is ge 2 6 18 ; then |
| 18 |
env_setup_xmakeopts |
| 19 |
emake headers_install |
| 20 |
INSTALL_HDR_PATH="${D}"/${ddir}/.. ${xmakeopts} || die |
| 21 |
|
| 22 |
In other words, for versions previous to 2.6.18, (almost) all headers |
| 23 |
in the tarball were installed. For 2.6.18 and later, the makefile in |
| 24 |
the tarball defines what gets installed. |
| 25 |
|
| 26 |
If you truly need this header, make a private copy for your source |
| 27 |
directory, or just define PID_MAX_DEFAULT yourself. But as I |
| 28 |
mentioned above, it can be changed at run-time, so relying on this to |
| 29 |
be constant could be dangerous. |
| 30 |
|
| 31 |
-Richard |
| 32 |
-- |
| 33 |
gentoo-user@g.o mailing list |
Archives