1:N threading

[Julian Elischer]

On Fri, 4 Apr 2003, Julian Elischer wrote:

> 
> 
> On Fri, 4 Apr 2003, Peter Wemm wrote:
> 
> > Daniel Eischen wrote:
> > > On Thu, 3 Apr 2003, Peter Wemm wrote:
> > > 
> > > > Daniel Eischen wrote:
> > > > 
> > > > > The patches are available:
> > > > > 
> > > > >   http://people.freebsd.org/~deischen/libpthread.diffs
> > > > > 
> > > > > FYI, since this is a new mailing list, the above changes
> > > > > are meant to give libpthread M:N capability.
> > > > > 
> > > > > I don't need testers; I have enough bugs that I know about
> > > > > to fix.
> > > > 
> > > > +	__asm__("movl %%gs, %0" : "=r" (id));
> > > > +	id >>= 3;
> > > > +	if (id - NLDT < 0)
> > > > 
> > > > There is a problem here, NLDT is kernel private and changes depending on
> > > > things like whether SMP is enabled or what the maximum number of cpus
> > > > is.
> > > > 
> > > > You're trying to find if its a local or global selector, right?
> > > > What you really want is bit 2 which tells you which it is.
> > > > 
> > > > #define ISLDT(s)        ((s)&SEL_LDT)   /* is it local or global */
> > > > #define SEL_LDT 4               /* local descriptor table */
> > > 
> > > OK, but if NLDT is kernel private, how do can I know
> > > what LDTs I can use as local?
> > 
> > Whoops.  I misread NLDT as NGDT.
> > 
> > Gah.  We're doing some crufty stuff here.  For starters, we're running
> > userland on a LDT for %cs and %ds/es/ss/etc.  We really should be using a
> > GDT slot for those.  Most of the other stuff there is for the a.out
> > "lcall 7,0" instruction and for BSDI's version of the lcall stuff.
> > 
> > The i386_[gs]et_ldt() syscalls really should have a way of reporting what
> > is available for use.
> 
> What we SHOULD be doing is setting an LDT entry to point to the 
> mailbox (puts some constraints on the mailbox but..) of each upcall
> and setting %gs to teh appropriate entry before returning to userspace.

by which I mean that the setting of the LDT would be done in the
kse_create() call by the kernel. The userland would just have to 
offset off the g segment to find the current KSE/upcall mailbox
which would have a pointer to the current thread..

> 
> (different schemes would be used for the other architectures.)
> 
> > 
> > Cheers,
> > -Peter
> > --
> > Peter Wemm - peter at wemm.org; peter at FreeBSD.org; peter at yahoo-inc.com
> > "All of this is for nothing if we don't go to the stars" - JMS/B5
> > 
> > _______________________________________________
> > freebsd-threads at freebsd.org mailing list
> > http://lists.freebsd.org/mailman/listinfo/freebsd-threads
> > To unsubscribe, send any mail to "freebsd-threads-unsubscribe at freebsd.org"
> > 
> 
> _______________________________________________
> freebsd-threads at freebsd.org mailing list
> http://lists.freebsd.org/mailman/listinfo/freebsd-threads
> To unsubscribe, send any mail to "freebsd-threads-unsubscribe at freebsd.org"
>