From: Michael Durrant (mdurrant@rt-control.com)
Date: Thu Feb 10 2000 - 09:32:32 EST
The uCsimm list has a short fuse for long messages. As it was intended for discussion only. I am cross posting the message to our uclinux-dev@uclinux.org list which does support larger file attachments. I have added Bernhard, Stuart and kwonsk to the list. -- Michael > From: "±Ç¼®±Ù" <kwonsk@mutech.co.kr> > To: <ucsimm@uClinux.com> > Subject: [uCsimm] RTL release 0.9J patch for uClinux (again) > Date: Thu, 10 Feb 2000 12:58:59 +0900 > X-Priority: 3 > X-MSMail-Priority: Normal > X-Mailer: Microsoft Outlook Express 4.72.3612.1700 > X-MimeOLE: Produced By Microsoft MimeOLE V4.72.3612.1700 > > (I' sorry if this mail send twice, strangly i received my message about > missing note and didn't received this message. therefore i resend this > msg again. Some mail problems?) > > ----------------------------------------- > > Hi, all and Bernhard > (Bernhard, i'm glad to here your voice again!) > > When i heard notes from Stuart Hughes and Burnhard, I was very happy. > because i felt that my porting seems a doable job. > ( During poring stages, i questioned myself that "Is this really doable job? > will this be useful for somebody? Who knows that this porting might be a > old timed no more used RTL API?". ) > > Push reset button - memory dumping -code searching - re arrange assembly > code > -another push reset button and again... > Yah, hard times is gone now. > > Here is a patch of my porting. And my note. > > I've used NMT-RTL release 0.9J as a base just because my desktop linix is > version 2.0.36 > with RTL release 0.9J I made this patch against fresh installed uClinux > 2.0.38. > If somebody need v1.0, well, i'll try. After seeing msg from Burnhard and > Stuart, i feel > that there might be small changes. because upper layer (rt_sche.c rt_time.c > rt_fifo.c) > was not touched more than 10 lines (i'm not sure, i didn't count it, may be > more than > 20 lines? i don't know right now :-) ) > > Diffrences from NMT-RTL's are > 1. rename moduel init/cleanup to xxx_init/xxx_cleanup and call this at > appropriate place. > rt_time_init, rtl_scheduler_init at main.c > and rtl_fifo_init at /drivers/char/mem.c -> this is because rt-fifo is > just another chardev. > and all chardev are initialized at there. > 2. add call to rtl_application_init at main.c when registering all chardev > driver(rt-fifo) was done. > -> this is because rt application might want to open rt-fifo. > 3. remove I8253_make_periodic functions (not make_task_periodic). -> this > is because > i386 plaform could use two timers (one channel for oneshot, another for > periodic) > but EZ328 has just one timer. Actually there are no reference to this > function > in the release 0.9J, nobody calls it ( i think) and periodic task just > needs oneshot timer. > 4. remove I8253_disable_speaker/enable_speaker. We don't have speaker :-) > 5. move RTL module's source tree to the arch/m68knommu/kernel/rtl > 6. move RTL module's header files to include/rtl > 7. As you know hardware specific part (interrupt controller and timer usage) > > Some code improvement for speed is possible. > 1. I just used stupid for loop for hooking RTaction this could be changed to > like > that of M68EZ328_do_irq. > 2. I've catched all interrupt at one handler, this could be changed to one > handler > for each interrupt level (0-7, 6?) Actually, this MUST be done because > if a RT task > which request interrupt for his own use, we must allow rt-preemption > with correct > interrupt priority level. I'll do this right now. > 3. Need many real testing of cource and timing analysis if somebody want it. > > Examples compilation note > before try to make frank example, make a symbolic link (for include search) > like this > ln -sf /opt/uClinux/linux/include/rtl /opt/uClinux/m68k-pic-coff/include/rtl > > That's all. > > BTW, could somebody post this patch on some anonymous ftp site for easy > access? > > Thanks > kwonsk@mutech.co.kr > > >
This archive was generated by hypermail 2.1.4 : Thu Sep 19 2002 - 13:19:14 EDT