[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[MiNT] analysing syscall.spp



Hi Konrad!

KK>I guess most are (if not all of them).
You can be sure that there are some calls which are not reentrant. I think
all calls which modify global data are not reentrant

KK>I think saving the context is necessary when a task switch is about to
KK>occur.
Well, the task switcher itself has to save the context of the current process
and after that to restore the context of the process to continue. But that's
normal business for a task switcher

The question is which calls modify more of the context than in normal TOS and
which calls modify the content of the context structure

KK>If you mean setting the in_kernel, all calls need that, because this
KK>flag indicates hum,.. that we are in kernel.
Well, my aim was the check of the in_kernel flag at the beginning of the
GEMDOS handler. Inbetween I had a look at syscall.spp.new, where this has
been removed and patched the kernel so that it now works with this new
syscall handler (I tell you more in a private mail)

KK>Hum... I don't know... when it is full? when context is changed? and
KK>probably few more conditions.
Well, since this is only one tst.w, when the buffer doesn't need to be
flushed, this is nothing of great interest to be removed

Bye

                Joerg