On 12/23/2014 05:39 AM, Peter Chen wrote: > On Mon, Dec 22, 2014 at 06:39:42PM +0530, Sanchayan Maity wrote: >> On 12/22/2014 06:48 AM, Peter Chen wrote: >>> On Fri, Dec 19, 2014 at 03:25:26PM +0530, Sanchayan Maity wrote: >>>> The first two patches add identification register API's. These can >>>> be used to get controller's revision. >>>> >>>> The third patch implements an errata for revision 2.40a. Not sure >>>> which other SOCs implement this version of the Chipidea core but >>>> this fixes the usb client issue observed on Vybrids. The patch was >>>> tested on a Toradex Colibri VF61 module with the 3.18 kernel. iperf >>>> tests ran for three hours plus, with these patches applied have found >>>> the USB client connection to be now reliable. >>> >>> Would you help do a overnight test? It is passed, I will queue them, >>> thanks. >> >> Yes definitely I can help with the testing. Are you looking for iperf >> tests only or something else? iperf tests running for 12 or 24 hours >> will do? I will need a bit of time to set things up here, as I am away >> from work, but, ya I will do them. >> > > iperf for g_ncm or g_ether and bonnie++ for g_mass_storage if you can > run, thanks. The tests were run on a Toradex Colibri Vybrid VF61 module having 256MB RAM with the 3.18 kernel. The iperf tests ran for around 19 hours before I stopped it. A snip of the iperf tests is below. Used the Ethernet Gadget class for this. [ 4] 70453.0-70453.5 sec 6.89 MBytes 116 Mbits/sec [ 4] 70453.5-70454.0 sec 6.83 MBytes 115 Mbits/sec [ 4] 70454.0-70454.5 sec 6.84 MBytes 115 Mbits/sec [ 4] 70454.5-70455.0 sec 6.89 MBytes 116 Mbits/sec [ 4] 70455.0-70455.5 sec 6.90 MBytes 116 Mbits/sec [ 4] 70455.5-70456.0 sec 6.90 MBytes 116 Mbits/sec [ 4] 70456.0-70456.5 sec 6.82 MBytes 114 Mbits/sec [ 4] 70456.5-70457.0 sec 6.80 MBytes 114 Mbits/sec [ 4] 70457.0-70457.5 sec 6.89 MBytes 116 Mbits/sec [ 4] 70457.5-70458.0 sec 6.85 MBytes 115 Mbits/sec [ 4] 70458.0-70458.5 sec 6.82 MBytes 114 Mbits/sec [ 4] 70458.5-70459.0 sec 6.82 MBytes 114 Mbits/sec [ 4] 0.0-70459.2 sec 946 GBytes 115 Mbits/sec Ran bonnie++ on gadget mass storage. CPU usage around the time of running this test was mostly around the 90% mark with the minimum at 60% plus. The storage directory was formatted with ext4. bonnie++ version used is 1.97 and was installed from the Arch repositories with pacman. The size of the file being specified for "lun" storage is 512MB. I have specified 128MB RAM in the below run with the size of file for IO performance as 256MB. Without this bonnie++ was giving me an error around the "Writing intelligently" point. I assume this has to do with the file size bonnie++ uses for testing. [sanchayan@Sanchayan-Arch ~]$ sudo /usr/bin/bonnie++ -m Vybrid -r 128 -d /var/run/media/sanchayan/Vybrid/ -x 5 -u root -n 0 -s 256 Using uid:0, gid:0. format_version,bonnie_version,name,concurrency,seed,file_size,io_chunk_size,putc,putc_cpu,put_block,put_block_cpu,rewrite,rewrite_cpu,getc,getc_cpu,get_block,get_block_cpu,seeks,seeks_cpu,num_files,max_size,min_size,num_dirs,file_chunk_size,seq_create,seq_create_cpu,seq_stat,seq_stat_cpu,seq_del,seq_del_cpu,ran_create,ran_create_cpu,ran_stat,ran_stat_cpu,ran_del,ran_del_cpu,putc_latency,put_block_latency,rewrite_latency,getc_latency,get_block_latency,seeks_latency,seq_create_latency,seq_stat_latency,seq_del_latency,ran_create_latency,ran_stat_latency,ran_del_latency Writing a byte at a time...done Writing intelligently...done Rewriting...done Reading a byte at a time...done Reading intelligently...done start 'em...done...done...done...done...done... 1.97,1.97,Vybrid,1,1419409300,256M,,659,87,8341,1,9401,0,4222,98,+++++,+++,3539,19,,,,,,,,,,,,,,,,,,23042us,66us,59us,4482us,79us,475us,,,,,, Writing a byte at a time...done Writing intelligently...done Rewriting...done Reading a byte at a time...done Reading intelligently...done start 'em...done...done...done...done...done... 1.97,1.97,Vybrid,1,1419409300,256M,,661,90,7689,1,9071,0,4011,99,+++++,+++,3426,20,,,,,,,,,,,,,,,,,,15406us,64us,62us,4667us,23us,10030us,,,,,, Writing a byte at a time...done Writing intelligently...done Rewriting...done Reading a byte at a time...done Reading intelligently...done start 'em...done...done...done...done...done... 1.97,1.97,Vybrid,1,1419409300,256M,,673,89,8117,1,9451,0,3879,98,+++++,+++,3355,22,,,,,,,,,,,,,,,,,,14210us,45us,69us,5069us,21us,10052us,,,,,, Writing a byte at a time...done Writing intelligently...done Rewriting...done Reading a byte at a time...done Reading intelligently...done start 'em...done...done...done...done...done... 1.97,1.97,Vybrid,1,1419409300,256M,,668,89,7801,1,9343,0,4099,98,+++++,+++,3336,16,,,,,,,,,,,,,,,,,,17019us,44us,75us,4920us,20us,10234us,,,,,, Writing a byte at a time...done Writing intelligently...done Rewriting...done Reading a byte at a time...done Reading intelligently...done start 'em...done...done...done...done...done... 1.97,1.97,Vybrid,1,1419409300,256M,,676,89,7953,1,9494,0,3878,98,+++++,+++,3396,22,,,,,,,,,,,,,,,,,,14080us,56us,42us,5177us,23us,12224us,,,,,, Let me know if this is OK. Thanks. -Regards, Sanchayan. > > Peter > >> -Regards, >> Sanchayan. >>> >>> Peter >>>> >>>> This patchset is based off on Shawn Guo's for-next branch >>>> https://git.kernel.org/cgit/linux/kernel/git/shawnguo/linux.git/tree/?h=for-next >>>> >>>> The credit for the patches and fix goes to Matthieu Castet and Peter Chen. >>>> First two patches are by Peter Chen and the third patch which fixed the >>>> bug we observed was reported by Matthieu Castet. >>>> >>>> The discussion of the problem and the relevant testing details can be found >>>> at this link: http://www.spinics.net/lists/linux-usb/msg118544.html >>>> >>>> The first version of this patchset originally send by Peter Chen can be >>>> found at this link: http://www.spinics.net/lists/linux-usb/msg118753.html >>>> >>>> Comments for review are welcome :). >>>> Note: I am going on a vacation so will not be able to reply or do any further >>>> tests till Monday. Will attend and take care of any comments/requests for >>>> further changes/testing from Tuesday. Apologize for the delay in advance. >>>> >>>> Sanchayan Maity (3): >>>> usb: chipidea: Add identification registers access APIs >>>> usb: chipidea: Add chipidea revision information >>>> usb: chipidea: Add errata for revision 2.40a >>>> >>>> drivers/usb/chipidea/bits.h | 10 ++++++++ >>>> drivers/usb/chipidea/ci.h | 53 +++++++++++++++++++++++++++++++++++++++++++ >>>> drivers/usb/chipidea/core.c | 23 +++++++++++++++++-- >>>> drivers/usb/chipidea/udc.c | 20 ++++++++++++++++ >>>> 4 files changed, 104 insertions(+), 2 deletions(-) >>>> >>>> -- >>>> 1.7.9.5 >>>> >>> > -- To unsubscribe from this list: send the line "unsubscribe linux-usb" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
