Difference between revisions of "Talk:BeagleBoard/DSP Clarification"

From eLinux.org
Jump to: navigation, search
(tidspbridge stability: add my signature)
(tidspbridge stability)
 
(2 intermediate revisions by the same user not shown)
Line 30: Line 30:
  
 
::::::: First of all, there's a difference between personal experience that is not shared with anybody else, and personal experience that is shared and confirmed (other members of my team can attest to dsp-bridge's unstability). Also, the fact that someone somewhere hasn't found a critical bug doesn't mean there are no critical bug, so the personal experience of someone who finds the dsp-bridge stable doesn't say much, the "personal" experience of someone who has found a bug *does* say a lot. It could be for example that you are being way too complacent with the dsp-bridge, while other people are stress-testing it. -- [[User:Felipec|Felipec]] 09:50, 1 June 2009 (UTC)
 
::::::: First of all, there's a difference between personal experience that is not shared with anybody else, and personal experience that is shared and confirmed (other members of my team can attest to dsp-bridge's unstability). Also, the fact that someone somewhere hasn't found a critical bug doesn't mean there are no critical bug, so the personal experience of someone who finds the dsp-bridge stable doesn't say much, the "personal" experience of someone who has found a bug *does* say a lot. It could be for example that you are being way too complacent with the dsp-bridge, while other people are stress-testing it. -- [[User:Felipec|Felipec]] 09:50, 1 June 2009 (UTC)
 +
 +
:::::::: Please report your findings, as a formal bug or steps to reproduce - [[User:Omar_rmz|Omar_rmz]] 14:30, 1 Jun 2009 (UTC)
  
 
== tidspbridge and pm ==
 
== tidspbridge and pm ==

Latest revision as of 07:39, 1 June 2009

Oldest implementation

dsp-gateway is not the oldest implementation, just the first one to be open source.--Jkridner 21:45, 6 May 2009 (UTC)

Right, I meant the oldest open source implementation, feel free to re-word. -- Felipec 00:05, 8 May 2009 (UTC)
I changed it to "It's the first open implementation and currently the more open" -- Felipec 23:14, 14 May 2009 (UTC)

tidspbridge stability

In my experience the DSP bridge is far from stable. I see multiple crashes and hangs, and I can even reboot the device by following a sequence of commands. Also proof of the instability is the multiple critical fixes that are being sent to linux-omap mailing list all the time. -- Felipec 10:16, 9 May 2009 (UTC)

Wikipedia and personal opinion "A stable version is one that has been through enough real-world testing to reasonably assume there are no showstopper problems, or at least that any problems are known and documented." but that does not mean that it won't benefit from any bug fix or enhancements. A fork from the stable version will be the code cleanup and redesign that is planned which might uncover some corner cases in bridge and open source modules as well. Again, any bug report is welcomed. -Omar_rmz 17:20, 9 May 2009 (UTC)

Exactly, there are showstopper problems therefore it's not stable. A bug is simply 'unexpected behavior' and there many kinds, some of the worst are; crashes and deadlocks, but the bridgedriver goes to an even worst level; it reboots the device. You can say it's a "corner-case" because you have seen it very rarely in your occasional testing, but when a device goes into production there would be thousands, perhaps millions of people, using it constantly with a much wider variety of clips. A corner-case that reboots the device is absolutely showstopper. -- Felipec 22:10, 9 May 2009 (UTC)
Please let me know when you report this as a formal bug or provide at least steps to reproduce. - Omar_rmz 00:35, 14 May 2009 (UTC)
I removed the comments regarding stability for now, both on dspbridge and mailbox/iommu. I have not seen TI saying that iommu/mailbox are not stable, please provide a reference. As for the dspbridge stability, if you don't continue the discussion here I'll add the comment again saying that it's unstable. -- Felipec 23:19, 14 May 2009 (UTC)
Again my personal opinion doesn't reflect a company, or are you speaking on behalf of Nokia? Look at my original comment "There are plans to share the mailbox and iommu that the dsp-gateway uses, whenever they are stable and have the full set of features supported by tidspbridge custom implementations, as well as to move parts of it to user-space." Do not rephrase and ask for references of things that you wrote. - Omar_rmz 00:35, 14 May 2009 (UTC)
I'm speaking as a member of the community. You said 'whenever they are stable' which means they are not stable yet, if TI considers them stable then there's no point in saying 'whenever they are stable'.
 ??? - I'll try to explain, 'whenever they are stable' means I tested them and there are failures and features missing, I'm preparing some patches on those... BTW I don't know if TI considers them to be stable or not. - Omar_rmz 6:53, 15 May 2009 (UTC)
I've never seen any comments from anyone regarding the stability of mailbox/iommu. On wiki pages it doesn't really make sense to write things out of personal experience, there needs to be a reference. -- Felipec 22:25, 15 May 2009 (UTC)
Just hilarious... read your first post on this discussion: "In my experience the DSP bridge is far from stable."; this is not leading any way, I'm fine if you think the driver is unstable, I'll try to keep working on it and your patches are welcomed to improve the code quality. Omar_rmz 23:53, 19 May 2009 (UTC)
First of all, there's a difference between personal experience that is not shared with anybody else, and personal experience that is shared and confirmed (other members of my team can attest to dsp-bridge's unstability). Also, the fact that someone somewhere hasn't found a critical bug doesn't mean there are no critical bug, so the personal experience of someone who finds the dsp-bridge stable doesn't say much, the "personal" experience of someone who has found a bug *does* say a lot. It could be for example that you are being way too complacent with the dsp-bridge, while other people are stress-testing it. -- Felipec 09:50, 1 June 2009 (UTC)
Please report your findings, as a formal bug or steps to reproduce - Omar_rmz 14:30, 1 Jun 2009 (UTC)

tidspbridge and pm

Omar: you added this "but now, after a lot of progress in pm branch, bridgedriver must be aligned with linux-omap-pm" there reference points to a master branch where there is no bridge driver, what is that supposed to prove? There are other branches in omapzoom that have the bridgedriver, I guess using a different pm than linux-omap. I'll remove that comment for now. Please provide a valid reference. -- Felipec 23:30, 14 May 2009 (UTC)

Agree on that, don't have a solid reference it was a personal opinion - Omar_rmz 00:47, 14 May 2009 (UTC)