Steve/BJ, I would like to re-confirm that USB is working with DMA enabled in your configurations.
The numbers that you have provided seems very very odd and we have never seen such low numbers in our testing on the TI kernels. On the GIT kernel capabalities TI has been pushing most of the enhancements to the GIT kernel through linux-omap GIT's. They should be available starting from the next rc (I think) as they are a part of Greg KH's queue of accepted patches. regards swami ________________________________________ From: bj [EMAIL PROTECTED] Sent: Thursday, November 13, 2008 1:57 AM To: Stephen Berry Cc: Subbrathnam, Swaminathan; [email protected] Subject: Re: USB patches (patch level 14) for LSP 1_20 I'm also just using the MV kernel instead of the git for the same reasons that Steve listed. I'm measuring write speeds in the order of 170KB/s. The VERY odd thing is that I've previously measured speeds in the multiple MB/s ranges. I'm seeing the exact same crippled write/read speeds in my SD writes as well which leads me to believe something else is going on with the system but I'm still very much interested on the progress of this patch. BJ Stephen Berry wrote: > Subbrathnam, Swaminathan wrote: > > Steve/BJ, > > Can you provide more details as to the issue you have reported. > What is the throughput you are seeing and under what configuration. > There was a new update that went out as I know of on DM355 that > should roll in the patch45 changes on DM644x w.r.t USB. Probably it > is yet to get posted to the external site. > > > On the update site there is the release kernel with one patch (14). > > I only have rough write numbers - since that is what we care about most > - approximately 600KB/sec. The same memory stick on my laptop does > > 4MB/sec. > > I assume you are discussing the issue in the context of the DM355 > GIT kernel. > > > I'd love to port to the GIT kernel, but my understanding is that it's > not ready for prime time yet (on the DM355). I've made quite a few > changes to get my CMOS sensor and decoder working in the current Ti/MV > kernel. > > If I'm wrong about the GIT kernel and it does everything that the MV > kernel does - please let me know and I'll take a closer look. > > regards > swami > ------------------------------------------------------------------------ > *From:* BJ Opp [EMAIL PROTECTED] > *Sent:* Wednesday, November 12, 2008 3:37 AM > *To:* Stephen Berry; [email protected]; > Subbrathnam, Swaminathan > *Subject:* RE: USB patches (patch level 14) for LSP 1_20 > > Steve, > > > > It seems you and I are chasing the same problems. I’m seeing pretty > horrid write speeds on USB and was getting ready to try and chase > down where they could be coming from so I second your question > regarding the patches. Has anybody seen/fixed this issue? > > > > BJ > > > > ------------------------------------------------------------------------ > > *From:* [EMAIL PROTECTED] > [mailto:[EMAIL PROTECTED] *On > Behalf Of *Stephen Berry > *Sent:* Tuesday, November 11, 2008 1:14 PM > *To:* [email protected]; [EMAIL PROTECTED] > *Subject:* USB patches (patch level 14) for LSP 1_20 > > > > > I'm running into a performance issue with using USB memory sticks - > namely very low write performance. I've recently ported to LSP 1_20 > (from the beta) and after searching the mailing list there seemed to > be some mention of a patch level 45 on the 6446 that would fix this > issue... > > But of course for the Dm355 distribution on the Ti update site, > patch level 14 is the current one. And to top it off, they don't > apply cleanly with quilt. > > So before I go through each patch individually I'd like to make sure > that these patches will help with my problem. Most of the comments > seem to be relating to OTG functionality - but the patch labeled > lsp_1_20_usb_intr_schedule_fix_014.patch - sounds like it could be > helpful. > > Any help is appreciated, > > Steve > > > > -- > Integral Design and Development > 312 Great Rd, Littleton MA 01460 > voice: 978 501 7658 fax: 603-679-3988 > > This e-mail and any files transmitted with it are confidential and intended > solely for the use of the individual or entity to which they are addressed. > If you have received this e-mail in error please notify the originator > immediately. The addressee is formally notified hereby that this e-mail and > any attachments are to be treated as confidential. > >_______________________________________________ Davinci-linux-open-source mailing list [email protected] http://linux.davincidsp.com/mailman/listinfo/davinci-linux-open-source
