Hi,
Attached is a linux unit with epoll system call wrappers. Also a demo is included that creates 100 pipes and writes a character to each and reads it non-blockingly.
Micha
Mailing lists / fpc-devel / 2006-02
Linux epoll syscalls + example
8 messages · 4 people · –
- Micha Nelissen 4
- Marco van de Voort 2
- Jonas Maebe 1
- Peter Vreman 1
Reply by email Subscribe to this list Your subscription
Read-only archive. Use j and k to move between messages.
Micha Nelissen wrote 5 lines
Hi,
Attached is a linux unit with epoll system call wrappers. Also a demo is
included that creates 100 pipes and writes a character to each and reads it
non-blockingly.
Marco, is it possible to update the sysnr.inc files in rtl/linux/<arch> from a newer version of the kernel ? It's probably less work than me trying to split all the syscall numbers up and you copy pasting them for each arch seperately?
Micha
Show quoted text 8 lines
> Attached is a linux unit with epoll system call wrappers. Also a demo is
> included that creates 100 pipes and writes a character to each and reads it
> non-blockingly.
Marco, is it possible to update the sysnr.inc files in rtl/linux/<arch>
from a newer version of the kernel ? It's probably less work than me
trying to split all the syscall numbers up and you copy pasting them
for each arch seperately?
First sb who is more into practical aspects of the Linux kernel should say which kernel we are targeting.
I don't even know which kernel I run on FC4
Marco van de Voort wrote 7 lines
> Marco, is it possible to update the sysnr.inc files in rtl/linux/<arch>
> from a newer version of the kernel ? It's probably less work than me
> trying to split all the syscall numbers up and you copy pasting them
> for each arch seperately?
First sb who is more into practical aspects of the Linux kernel should say
which kernel we are targeting.
Defining the syscall numbers is independent of choosing the target linux kernel, right ?
I don't even know which kernel I run on FC4
Probably 2.6.10 or so, look at "uname -a" ;-).
I think most will have a 2.6 kernel by now, except for people running odd architectures with lazy kernel maintainers.
Micha
Show quoted text 13 lines
On Fri, Feb 03, 2006 at 12:23:08PM +0100, Marco van de Voort wrote:
> > Marco, is it possible to update the sysnr.inc files in
> rtl/linux/<arch>
> > from a newer version of the kernel ? It's probably less work than me
> > trying to split all the syscall numbers up and you copy pasting them
> > for each arch seperately?
>
> First sb who is more into practical aspects of the Linux kernel should
> say
> which kernel we are targeting.
Defining the syscall numbers is independent of choosing the target linux
kernel, right ?
The kernel numbers should be taken from the latest kernel (2.6.15). They are backwards compatible.
Defining the syscall numbers is independent of choosing the target linux kernel, right ?…
Show quoted text 7 lines
On Fri, Feb 03, 2006 at 12:23:08PM +0100, Marco van de Voort wrote:
>
> First sb who is more into practical aspects of the Linux kernel should say
> which kernel we are targeting.
Defining the syscall numbers is independent of choosing the target linux
kernel, right ?
It should. But sometimes call names are changed to _old and new ones with the same name appear (with e.g. rt_ prefixed).
So I'm not going to commit changes to sysnr.inc without it at least having some testing, or evaluation of possible painpoints of running on older versions.
Show quoted text 6 lines
> I don't even know which kernel I run on FC4
Probably 2.6.10 or so, look at "uname -a" ;-).
I think most will have a 2.6 kernel by now, except for people running
odd architectures with lazy kernel maintainers.
IIRC FC4 comes standard with at least 2.4 as choice. Don't underestimate peopl that run under standard circumstances.
Marco van de Voort wrote 6 lines
> I think most will have a 2.6 kernel by now, except for people running
> odd architectures with lazy kernel maintainers.
IIRC FC4 comes standard with at least 2.4 as choice. Don't
underestimate
peopl that run under standard circumstances.
The standard kernel of both FC3 and FC4 is 2.6.x.
Jonas
It should. But sometimes call names are changed to _old and new ones with the same name…
Marco van de Voort wrote 9 lines
> Defining the syscall numbers is independent of choosing the target linux
> kernel, right ?
It should. But sometimes call names are changed to _old and new ones with
the same name appear (with e.g. rt_ prefixed).
So I'm not going to commit changes to sysnr.inc without it at least having
some testing, or evaluation of possible painpoints of running on older
versions.
Yes, but a diff will quickly resolve that: simply only add names for the numbers you do not have yet, and do not remove previous ones.
Micha