This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

AM3352: How to solve the problem of low transmission efficiency after DMA is enabled on USB?

Part Number: AM3352

hi  ti

     The kernel version we are using now is 3.2. The problem is that we have a 4G Cat4 cellular module connected to the USB 0. When we use Speedtest to measure the cellular rate, the speed measurement will fail (the downlink measurement will fail). By default, USB DMA transmission is not enabled in version 3.2. We found that the module transmits a large number of data packets in a short time, and the CPU takes a lot of time to read data in the USB interrupt, As a result, the soft interrupt of the kernel network protocol stack has no chance to execute, and the read data is overstocked (there may be nearly 300 messages overstocked), resulting in the failure of speed measurement. After enabling USB DMA transmission, we found that the speed measurement was successful, but the uplink speed could not go up. After debugging, we found that there was a problem with the USB DMA driver and the transmission efficiency was not high, so the speed could not go up;

    1. We browsed the Ti E2E, and some of the answers said that it was a silicon bug. It is not clear what version has been updated. Currently, the kernel version we are using is 3.2. We expect to repair it in this version. Is there any method?

    2. Can you provide information about the development of the chip am3352 and cppi41? pls send my email:  rain.wang@arrow.com

    3. I tried to download the file with this link below, but it failed. It prompted me that I have no permission. If you can view it, please transfer it to me. Thank you.

         e2e.ti.com/.../8547.sprz360e_5B00_1_5D00_.pdf

  • Hi Rain,

    The kernel v3.2 for AM335x was released many years ago, and it is no longer supported.

  • Hi Bin

         1. Which version solves the above problems?

         2. Can the information required above be provided?

  • Hi Rain,

    The symptoms you described won't tell what the root cause is in the kernel drivers, It would require further debug to understand the issue.

    If the issue also happens in the latest Processor SDK, we will help to debug it.

             e2e.ti.com/.../8547.sprz360e_5B00_1_5D00_.pdf

    This is not a link, so I am not sure what the document is. Is it possible to provide the link of the pdf file?

  • hi  bin

    1. http://e2e.ti.com/cfs-file.ashx/__key/communityserver-discussions-components-files/790/8547.sprz360e_5B00_1_5D00_.pdf

    2. Can you provide information about the development of the chip am3352 and cppi41? pls send my email:  rain.wang@arrow.com

    3. I have asked the customer to try to upgrade to the latest version 5. X SDK on the official website, but the customer needs to spend some time. Do you have any application notes for upgrading the kernel to share?

  • add:Also, can you share the description data of USB driver? It is mainly about the data of cppi41 and ti81x, and the definition of the registers in the driver. The customer can't get the specific functions without the data.

  • Hi Rain,

    1. e2e.ti.com/.../quote]

    Do you have the link to the e2e thread which refer to this PDF file? That might give a clue about this PDF.

    2. Can you provide information about the development of the chip am3352 and cppi41?

    CPPI41 is documented in the AM335x TRM. What else information do you need?

    Do you have any application notes for upgrading the kernel to share?

    Sorry we don't have such type of migration guide.

    can you share the description data of USB driver? It is mainly about the data of cppi41 and ti81x, and the definition of the registers in the driver

    I am not sure what do you mean by "description data". The CPPI41 description and registers are in the TRM. The MUSB core registers are in a NDA document (this is not a typical NDA with TI, rather a special NDA dedicated to this MUSB controller).

    [/quote]