l4-hackers January 2003

l4-hackers@os.inf.tu-dresden.de

9 participants
7 discussions

RE: Deterministic scheduling with local APIC
by Jacob Gorm Hansen 05 Feb '03

05 Feb '03

On Mon, 2003-01-13 at 16:06, Josh English wrote: > Jacob, Josh, I hope you don't mind me CCing this to the L4 list. > The largest problem I see with this scheme is that I cannot > conceive of a mechanism where the data sent to the CPU and > the failover CPU will not cause enormous delays. This is > because any failover system will require a confirmation of > the data received by the backup CPU. Even if the transport > and confirmation mechanism was hardware, the bottleneck > inherent in the design would render the system unfeasible in > a real-world implementation. I imagine that if the system is only ever accessed from the network, all the data that is to be duplicated comes from there. I suppose one could just use an Ethernet hub or similar hardware to make sure all Ethernet frames arrive at both hosts, coupled with a dedicated low-latency link between the hosts used for syncronisation (which given certain guarantees about the performance of the local Ethernet segment could be as simple as just each host counting all incoming frames and sharing the counter via the dedicated link). > Since my understanding of deterministic scheduling is that > in multiplicity scheduling, jobs are partitioned into > job types where the jobs in each type are identical. Then > the job is sent to any given machine as determined by the > scheduling mechanism. Thus sending every job to every > machine would cause much greater overhead than is justified > by having a failover mechanism. What I meant by 'deterministic' is just that all threads always get their full time slice. For this to happen a means of determining upon the arrival of an interrupt exactly how many cycles the running thread has left of its slice, and a means of granting it precisely this number of cycles before it gets preempted, is needed. I suppose you could use the TSC for the former and perhaps the local APIC for the latter, but I am unsure if trying to do so would be crazy. > Well, unless you are building systems to withstand nuclear > attack or something. So, if you want to write the code, I > think that your system would work provided your 'master' > scheduling mechanism resides above the kernel scheduler. > I think that this would prevent race conditions from > forming becuase hardware delays would be invisible to the > kernel at that point. I don't know if I will ever find time to write something like this, even if it were possible, but it struck me as an interesting future project. best, Jacob

2 2

Fiasco compiling problems
by Manel Jimenez 29 Jan '03

29 Jan '03

Hi!I'm trying to build fiasco kernel but i've run in some troubles. After "make hello" it gives mes this error: ../../../../../lib/x86_586/l4v2/libl4dbg.a(i386-stub.o)(.text+0xac3): In function `getpacket': /home/juan/os/l4/pkg/l4dbg/lib/src/OBJ-x86_586-l4v2/../i386-stub.c:526: undefined reference to `strlen' collect2: ld returned 1 exit status In this file(i386-stub.c) "#include <string.h>" was commented, so i uncommented but the problem still persists.What am I doing wrong?Has anybody got this problem? Thanks in advance __________________________________________________ Do you Yahoo!? Yahoo! Mail Plus - Powerful. Affordable. Sign up now. http://mailplus.yahoo.com

2 1

User space scheduling, 2nd try
by Robert Kaiser 24 Jan '03

24 Jan '03

Hi I am currently pondering about ways to implement time partitioning in an L4 context, i.e. the idea is to distribute processing time to subsystems (consisting of groups of tasks/threads) according to predefined rules. Ideally, the underlying subsystems should be unaware of this process. Also, ideally, the concept should be recursive (i.e. a subsystem task should be able to donate (some of) its processing time to other sub-subsytems). At first glance, the concept of external preempters, as specified (though AFAIK never implemented) in L4 looks like it could be a useable basis for this, however: - I'm still having difficulties to fully understand preempters in L4. Their specification in the reference manual is somewhat sketchy. - I have a few ideas of my own, but implementing them would require extensions to some L4 functions. My past experience with the L4 reference manual has been that everything in there is there for a good reason, but the reasoning is not always explained (which is probably the way a "reference manual" should be). I would very much appreciate if someone could point me to a document that explains the reasoning behind preempters, as defined in L4. Failing that, are there any publications on external schedulers in the context of microkernels that might be helpful to read, any discussion forums you might suggest? .. or am I really on uncharted ground here? Any response would be greatly appreciated Rob ---------------------------------------------------------------- Robert Kaiser email: rkaiser(a)sysgo.de SYSGO AG Am Pfaffenstein 14 phone: (49) 6136 9948-762 D-55270 Klein-Winternheim / Germany fax: (49) 6136 9948-10

2 2

User space scheduling
by Robert Kaiser 21 Jan '03

21 Jan '03

Hi experts, I am trying to identify (and examine) ways to implement user-level scheduling on L4. I read in Michael Hohmuth's diploma thesis on L4Linux that it would have been desirable to use the preemption handlers to implement Linux's scheduling policy, but since L4 did not support them at the time, he used L4's built-in scheduler instead. Also I read a posting by Gernot Heiser in the mail archive of this list in which he suggests that the preemption handler concept is somehow flawed. To my knowledge, none of the current L4 variants has implemented the concept. To be honest, I'm having difficulty to fully understand the concept. For instance, the difference between an internal and an external preempter is not clear to me. So my questions are: - is there any further reading (besides the L4 Reference Manual and the L4 User Manual) that describes the concept ? - has the concept actually been put to the test ? - roughly, how would I implement a specific scheduling policy in e.g. L4Linux based on this concept (assuming L4 does provide it) ? Thanks in advance for any insights! Rob ---------------------------------------------------------------- Robert Kaiser email: rkaiser(a)sysgo.de SYSGO AG Am Pfaffenstein 14 phone: (49) 6136 9948-762 D-55270 Klein-Winternheim / Germany fax: (49) 6136 9948-10

1 0

Deterministic scheduling with local APIC
by Jacob Gorm Hansen 13 Jan '03

13 Jan '03

hi again, [OT: First of all a great thanks to everyone at TU Dresden for the interesting discussions and pleasant company during our visit last week.] Following a discussion with Jan Glauber Friday evening on the HP Somersault API, which does failover by making sure processes on a standby machine receive the exact same inputs as those on the master machine, I wondered if anyone knows if the PentiumII local APIC is precise enough that you can do completely deterministic scheduling, say, in Fiasco? It seems to me that one could implement an OS-level failover system by wrapping the OS in an I/O environment (like NomadBIOS ;-)) which would somehow ensure that the exact same input was received at both the master and the slave hosts, provided that the scheduling performed by the kernel scheduler could be performed with clockcycle precision (otherwise race conditions would allow things to get out of sync). Does anyone have an opinion as to whether or not this might be possible using current Intel hardware? best, Jacob

1 0

looking for PCA-200E firmware&driver
by uaca＠alumni.uv.es 09 Jan '03

09 Jan '03

Hi I'm interested in download the source of the Fore-PCA200 firmware the DROPs project made (source code), and also the driver I haven't found that on the guest access to the cvs server, etc... I have tried to contact Uwe Dannowski and other L4-ATM people with no luck anybody can help me? Thanks in advance Ulisses Debian GNU/Linux: a dream come true ----------------------------------------------------------------------------- "Computers are useless. They can only give answers." Pablo Picasso ---> Visita http://www.valux.org/ para saber acerca de la <--- ---> Asociación Valenciana de Usuarios de Linux <---

3 3

how current l4(with oskit) implement the driver framework?
by Yu Neil 07 Jan '03

07 Jan '03

hello: how current l4(with oskit) implement the driver framework? ( the oskit`s ? ), and where is the Omega0 server ? ( not implement ? ) Neil Yu _________________________________________________________________ 享用世界上最大的电子邮件系统― MSN Hotmail。 http://www.hotmail.com

1 0

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

2002

2001

2000

1999

1998

l4-hackers January 2003