This is the mail archive of the libc-alpha@sources.redhat.com mailing list for the glibc project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: DLOPEN and Statically linked executable


> The question is not in having some special static thread stack size by
> default, but rather possibility to dynamically select it.  Currently
> even if you set thread stack to smaller value it will still be
> allocated on 2M boundaries not allowing to have that much number of
> threads.

I know, I was arguing that the maximum stack size should remain at
least 2M.

> Also I would not agree about such architecture.  Thread-per client
> architecture is quite convenient for many applications.   Not only
> MySQL but various other applications use it (ie OOPS). With only
> change it is process per client Apache 1.3.x is using this model.

apche spawns one process per client?  For all I know, this is simply
false; I'd argue that to be a severe bug, too.  AFAIK, all forked
clients process a configurable amount of connections, typically
hundreds.

> Any hard coded limits are not good, it does not matter if it is number
> of users, processes, threads or open files.

Sure, but the hard-coded limit in this case is the ia32 3GB address
space.  It's a tradeoff: you can have either reasonably large stacks
or a huge amount of threads.  Since I don't have hundreds of CPUs, I'd
always choose the reasonable stack size.

Regards,
Wolfram.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]