- ixp1200 - lists.cs.princeton.edu

Re: [ixp1200] waiting for memory writes
by Shyamal Pandya 18 Nov '02

18 Nov '02

Use the sram/sdram instruction with a sig_done optional token. This won't swap out the context. A wait loop can then be written using the br_!signal instruction. Refer the microcode programmer's reference for details. It even has an example. All this if you are using microcode. >From: Gregory Prier <greg(a)Lever.cs.ucla.edu> >Reply-To: ixp1200(a)CS.Princeton.EDU >To: ixp mailing list <ixp1200(a)CS.Princeton.EDU> >Subject: [ixp1200] waiting for memory writes >Date: Mon, 18 Nov 2002 15:31:26 -0800 (PST) > >Is there any way to get the uengine to wait for a memory write to complete >without swapping context? I can swap context and wait for a signal but >then >other threads have a chance to run, which I don't want to happen. I just >want >the processor to spin and waste cycles while the write completes. Yes, I >know >this is inefficient, but it would be a big help right now for debugging. > >Thanks for the help. >GKP > >---------------------------------------------------------------------- >Greg Prier Previously: A Mudder >3564 Boelter Hall Currently: A Bruin working >UCLA CA 90095 toward a PhD >(310) 825-8899 After that: No idea > >"Hell, there are no rules here. We're trying to accomplish something." > Thomas Edison _________________________________________________________________ Add photos to your messages with MSN 8. Get 2 months FREE*. http://join.msn.com/?page=features/featuredemail

1 0

waiting for memory writes
by Gregory Prier 18 Nov '02

18 Nov '02

Is there any way to get the uengine to wait for a memory write to complete without swapping context? I can swap context and wait for a signal but then other threads have a chance to run, which I don't want to happen. I just want the processor to spin and waste cycles while the write completes. Yes, I know this is inefficient, but it would be a big help right now for debugging. Thanks for the help. GKP ---------------------------------------------------------------------- Greg Prier Previously: A Mudder 3564 Boelter Hall Currently: A Bruin working UCLA CA 90095 toward a PhD (310) 825-8899 After that: No idea "Hell, there are no rules here. We're trying to accomplish something." Thomas Edison

1 0

SOS, a question about connection and output!
by Ye, Irvin 18 Nov '02

18 Nov '02

Hi, I'm puzzled on the connection for ixp1200. I had tried to connect PCs and ixp1200 ports in standard way: via a hub / switch, but it failed to work. When I tried direct connection through crossover twisted pairs, it seemed OK, packets could be successfully sent to ixp1200 ports. But then it found that packets could not be sent from the ports. When I tried Intel's simple bridge example, I found there were problem in the egress side: packets could not be sent out, so finally there was a RmSendPacket error. Upon Erik's suggestion, I've checked some status of the board and found that when there were packets to send, the value of the XMIT_PTR regisger would stick to a fix number, and the active transmit thread would cycle around these codes: Egress_WaitForPortReady[abs_port] ctx_arb[sdram] I've tried long, but failed to resolve the problem. It's making me crazy. Who can give me a help? Regards. Irvin

2 1

Prefetch and Non-Prefetch SDRAM
by Shyamal Pandya 18 Nov '02

18 Nov '02

Hi, Could someone tell me the distinction between prefetch and non-prefetch SDRAM on the IXP1200? And why does the resource manager provided with the IXA Ace framework use only the uncached SDRAM? While at it could someone tell me what is the cache flush area meant to be? Thanks! Shyamal Pandya. Arizona State University. _________________________________________________________________ Help STOP SPAM with the new MSN 8 and get 2 months FREE* http://join.msn.com/?page=features/junkmail

2 1

recycling NICs from IXP12EB's, gig NICs & multiple IXP12EB's in a PC (running VERA)
by Andrew Moore 18 Nov '02

18 Nov '02

Q1. I have vera working happily on an ixp12eb (rev C.), can the ethernet board that comes with the ix12eb kit be recycled into a linux box? I'm tempted to try, and did in fact do this, but the machine wasn't happy booting so I figured it could be either the ancient machine I was trying or something more fundamental in the ethernet nic that comes with the ixp12eb. Q2. I'm interested in talking to people that have used vera and the 1 Gig interfaces... anyone out there? I'm interested in using them for traffic generation, but would like to talk to anyone who has done more than the basic "turn it on" and "yep the physical layer is working" test. Q3 now this one might be a bit odd, however.... What problems are there with sticking several ixp12eb's into a machine? I assume in its current incarnation the vera drivers are not aware of several boards? (no I've not tried it) any ideas what the load is for an ixp12eb? is it going to 'cause the mobo pci supplies to be draining too much? This would be for a "lots of ports" traffic generator.... although a "lots of ports" might be good too. cheers, Andrew. Andrew Moore, Computer Laboratory, University of Cambridge, Ph: +44-1223-763446 Fax: +44-1223-334678 andrew.moore(a)cl.cam.ac.uk www.cl.cam.ac.uk/users/awm22/

2 1

princeton roms on a rev Bixp1200 Eth. Eval. Board.
by Andrew Moore 18 Nov '02

18 Nov '02

Hi, I've been attempting to get VERA up on a board with rev B arm core (at least its what I had always thought was a REV B - this may be the problem) Possibly there are many more versions of the IXP1200 Ethernet Evaluation Board than I realised and I may be stuck with one that will never fly... I've been following the "update the flash" instructions (http://www.cs.princeton.edu/nsg/vera/README.flash.txt) However everything is fine until futil, Instead of getting C:\>FUtil -c 1 -n -a 0 -o 0 -e 200000 -f FILE.bin Flash Utility, Version 1.2.76 FlashUtil, Version 1.2.76 CPU Revision: 6901C123 Flash Width: 32 Flash Bank Size: 2048 KBytes Flash Type Bank[0]: 28F800B3-B Flash Type Bank[1]: 28F160B3-T Flash Type Bank[2]: 28F160B3-T Flash Type Bank[3]: 28F160B3-T Sending data to remote system... Percent complete: 100 Updating flash... Percent Complete: 100 done I get C:\> FUtil -c 1 -n -a 0 -o 0 -e 200000 -f flashImage.bin Flash Utility, Version 2.01.112 FlashUtil, Version 1.1.109 CPU Revision: 6901C121 Flash Width: 32 Flash Bank Size: 2048 KBytes Flash Type Bank[0]: 28F800B3-B Flash Type Bank[1]: 28F160B3-T Flash Type Bank[2]: Unknown Flash Type Bank[3]: Unknown Sending data to remote system... Percent complete: 100 Error in doing inflate: 'Error FFFFFFFD C:\> Clearly this board doesn't have 2Mb of flash, or its not in the configuration expected. I've tried the same process on other IXP ether Eval boards and it works, but not this particular type. Details: the boards I have that work call themselves Rev C the board has the identifier A13033-001 A02 L1 (this is from the sth/wst corner of the board) (near serial port) the board that gives the error above identifies itself as a Rev B and has the identifier A13033-002 Is this board a waste of time??? Thanks for your thoughts, Andrew. Andrew Moore, Computer Laboratory, University of Cambridge, Ph: +44-1223-763446 Fax: +44-1223-334678 andrew.moore(a)cl.cam.ac.uk www.cl.cam.ac.uk/users/awm22/

6 8

IXP1200 - RHL
by Rayashish＠aol.com 15 Nov '02

15 Nov '02

Hi all, Can anyone help me to load Linux2.4 kernel on IXP12eb sitting on host pc running RHL2.4. --I have VERA drivers loaded on Linux Host. --Can connect to ixp-board through Hyper teminal. --Cygmon is running in flash. --Also Vera loader is running in flash which waits for start address of srec-image. How do i load arm-kernel and ramdisk image through host pci usinng vera tools? Any help or pointer- TIX --Ashish

2 1

RE: [ixp1200] IXP1200-2505 PCI card booting
by Brandewie, Dirk J 15 Nov '02

15 Nov '02

The problem here is that the card saw a space bar that told it to drop out of loop where it was waiting for the kernel to be downloaded and go to cygmon. The ixstart script waits to see the login prompt from the kernel booting on the board. The script times out if it does not see the login prompt. If you reboot the system and connect to the serial port that the board is connected to you should see the message: "Press space bar to abort host download mode...." repeated over and over. If this is the case you should be able to exit minicom and use the bootixp script to boot the board. --Dirk > -----Original Message----- > From: Ning Weng [mailto:nweng@ecs.umass.edu] > Sent: Wednesday, November 13, 2002 3:45 PM > To: ixp1200(a)CS.Princeton.EDU > Subject: [ixp1200] IXP1200-2505 PCI card booting > > > > Hello, > > Sorry for bother but I really need some kind inputs... > > Did I any one have this kind of experience? Plese help and give me > some hints. > > For IXP1200 ENP-2505 PCI card, following the IXA Education > Workstation > Setup Guide, I did successfully boot the card with the > message" Finished > ENP-2505 Driver Install..." . But after a few days, when I > reboot, this is > what I got : > > ".... > > Welcome to minicom 1.83.1 > > OPTIONS: History Buffer, F-key Macros, Search History Buffer, I18n > Compiled on Aug 28 2001, 15:09:33. > > Press CTRL-A Z for help on special keys > > > cygmon> script "./.ixstart": global timeout > > ...." > > I cann't remember what I exactly did after my first > successful booting. > But I try to reinstall the software and hardware, but I got the same > message. > > Any help is greatly appreciated!!!! > > Thanks a lot. > > -Ning >

2 2

microengine loader (w/correct URL)
by Kenneth M. Mackenzie 13 Nov '02

13 Nov '02

[doh -- always check the URLs] Hi all, Here's an updated package of our local tools, just updated to include a microcode loader, etc. that we use to operate the microengines from the StrongARM under Linux without ACEs or the workbench. http://www.cercs.gatech.edu/projects/asan/software.html Note that the StrongARM tools (in the "utils/" directory) don't depend on our alternative host-side driver scheme, I'm just packaging them in one tarball. ---------------- Various folks have asked about alternatives to ACEs. We use three schemes to program the microengines: 1. Homebrew tools including loader and etc., that run stand-alone under Linux on the StrongARM and communicate with the uEngines via Intel's "ue.o" kernel module on the StrongARM (same as used by rs_udebug with the workbench). We also use command-line versions of the assembler and linker in this configuration under "wine" so it's an all-linux solution. 2. Intel's Workbench. You can use this to load code onto the hardware instead of the simulator and it doesn't require ACEs (in fact it doesn't work with ACEs although it's not obvious that they should be incompatible). 3. ACEs Note: I would not characterize any of these three toolchains as exactly "easy". Each just offers its own kind of pain. In other words, don't switch toolchains just because you're frustrated with ACEs. -- Ken

2 1

VERA LibIXPRouter Minor Software Update
by Yitzchak M. Gottlieb 13 Nov '02

13 Nov '02

We are very happy to announce an update of the Vera IXPRouter library. This is a very minor update to correct small problems in compiling the IXPRouter library. Some relevant changes include: 1) Removed references to atoi in Libixprouter/sa/src/control.c (I applied the patch previously posted to the mailing list.) 2) Removed reference to p2tosa.h in Libixprouter/common/src/Makefile The software is available at http://www.cs.princeton.edu/nsg/vera. Please feel free to post any questions. Zuki -- Yitzchak Gottlieb zuki(a)CS.Princeton.EDU

1 0