Date: Tue, 15 Mar 2022 12:57:26 +0800 From: Archimedes Gaviola <archimedes.gaviola@gmail.com> To: Hans Petter Selasky <hps@selasky.org> Cc: freebsd-arm@freebsd.org Subject: Re: Raspberry Pi 3B USB Printing Issue Message-ID: <CAJFbk7G_SVcoH4pak4SvNp0EYsbU_i4LiNZ6HyEDrmxAPvviUg@mail.gmail.com> In-Reply-To: <CAJFbk7EutGbB-LoHxN_mUZbSb70C8yjxfZBADvejuCLwTcD5FA@mail.gmail.com> References: <CAJFbk7EzSfPNbaGxiweKrivwNrKXCPVzA1b7_=0_bTvbs8oBow@mail.gmail.com> <7c67118e-f6ec-c87d-9a81-3ee6a5952f49@selasky.org> <CAJFbk7GYbLAFTJY077Nzh3CTBJM6bk8swr4AkgGMaukCxrfcHQ@mail.gmail.com> <60f98b10-dcdc-cdf4-3d7a-fe9fd4dff223@selasky.org> <CAJFbk7EAjrQG5Kj_upVKW72opOS%2B8d63VrMnQdLxcJjUcfsd=g@mail.gmail.com> <8226461b-5740-9c19-0575-2740bd952e16@selasky.org> <CAJFbk7GwjFA-=GrJG3KTCnqVfEPhRSY1g8xyws_nE8pAohErEg@mail.gmail.com> <dabb798c-435c-6dd3-ac9b-8db3fb02a43c@selasky.org> <CAJFbk7FFeNTKvbNMr41kkwwYtyamJybTzk3=DQB1Hg3z%2Bx2hgQ@mail.gmail.com> <CAJFbk7GZF4ahORzCUKaLZS4b=fCJQbEADPJmMYZjyaJwRR%2Bhbw@mail.gmail.com> <CAJFbk7Fxz24r73AzHGhZHHKYwyHLqn-HCpiCpsAi4D-E6oKtKg@mail.gmail.com> <7a77e3bd-1186-56a6-e60e-89e51c190a01@selasky.org> <CAJFbk7EutGbB-LoHxN_mUZbSb70C8yjxfZBADvejuCLwTcD5FA@mail.gmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
--000000000000c7449305da3aa2b0 Content-Type: text/plain; charset="UTF-8" On Mon, Mar 14, 2022 at 6:12 PM Archimedes Gaviola < archimedes.gaviola@gmail.com> wrote: > > > On Mon, Mar 14, 2022 at 5:31 PM Hans Petter Selasky <hps@selasky.org> > wrote: > >> On 3/14/22 10:20, Archimedes Gaviola wrote: >> > On Sun, Mar 13, 2022 at 11:25 PM Archimedes Gaviola < >> > archimedes.gaviola@gmail.com> wrote: >> > >> >> >> >> >> >> On Sun, Mar 13, 2022 at 2:27 PM Archimedes Gaviola < >> >> archimedes.gaviola@gmail.com> wrote: >> >> >> >>> >> >>> >> >>> On Sun, Mar 13, 2022 at 12:38 AM Hans Petter Selasky <hps@selasky.org >> > >> >>> wrote: >> >>> >> >>>> On 3/12/22 16:31, Archimedes Gaviola wrote: >> >>>>>> >> >>>>>> IOERROR usually means an electrical error. The RPI3B will use a >> >>>>>> transaction translator for the FULL speed traffic, which is driven >> by >> >>>>>> software. >> >>>>> >> >>>> >> >>>> Hi Archimedes, >> >>>> >> >>>>> Hi Hans, >> >>>>> >> >>>>> I'm curious about the transaction translator you've mentioned, any >> >>>> idea why >> >>>>> there's a need for translation and what is being translated? >> >>>> >> >>>> When the High Speed USB HUB was invented, there was a need to support >> >>>> FULL and LOW speed USB transactions. Because FULL and LOW speed >> >>>> transactions are slow and take up much bandwidth, a transactions >> >>>> translator was invented which translates between High Speed USB and >> >>>> FULL/LOW speed USB. That means the RPi 3B consists of a single USB >> HIGH >> >>>> speed port, followed by a USB HUB. These transactions are not >> visible in >> >>>> usbdump . >> >>>> >> >>>> >Does this only >> >>>>> happen when RPi 3B (acting as host controller) is transmitting data >> to >> >>>> the >> >>>>> Epson printer? Are translation events visible in the usbdump? In >> this >> >>>> case >> >>>>> there's a way to possibly track what's going on and how to identify >> any >> >>>>> info that is being translated? >> >>>> >> >>>> By turning on the HC debugging, you can possibly track via debug >> >>>> messages what is going on. Maybe it is a timing issue, that the SW is >> >>>> too slow serving the micro transactions. >> >>>> >> >>>> Any idea also if translation is being >> >>>>> performed in RPi 4B? >> >>>> >> >>>> The later RPi's do the transaction translator bits in HW or via the >> XHCI >> >>>> block. >> >>>> >> >>>> As I have conducted several printing cases with my >> >>>>> Epson printer without any issues with either of the 4 ports. >> >>>>> >> >>>>> Sorry for these questions as I am catching-up to the USB technology >> >>>>> internal workings under the hood. >> >>>> >> >>>> You are welcome! >> >>>> >> >>> >> >>> >> >>> Thank you so much Hans for answering my questions, really appreciate >> it! >> >>> I have a much better understanding now. >> >>> >> >>> I came from testing with 13.0-RELEASE having the same RPi 3B hardware >> and >> >>> setup and it's very stable. I haven't encountered any IOERROR and >> >>> incomplete printed outputs that were encountered with 14.0-CURRENT >> >>> (February 24, 2022). By the way, I found in the repository here >> >>> https://download.freebsd.org/snapshots/arm64/aarch64/ISO-IMAGES/14.0/ >> >>> that there were two latest snapshots released dated March 3 and March >> 10, >> >>> respectively. I need to take printing tests first, especially the >> latest to >> >>> check if it also manifests before I go back to the Feb. 24 snapshot >> and do >> >>> a thorough test with debugging. I'll provide updates for any >> observations. >> >>> >> >>> Thanks, >> >>> Archimedes >> >>> >> >> >> >> >> >> Hi Hans, >> >> >> >> Initial testing conducted with the latest 14.0-CURRENT (March 10, 2022 >> >> snapshot) seems to be stable. Another test will be performed tomorrow. >> >> >> >> Thanks, >> >> Archimedes >> >> >> > >> > Hi Hans, >> > >> > I still encountered the issue this morning with 14.0-CURRENT (March 10, >> > 2022 snapshot). I just noticed when I left my RPi 3B idle for a long >> period >> > of time (2-3 hours and more) and started printing, then the issue >> pops-up. >> > Another concern I encountered was when I started enabling >> > hw.usb.dwc_otg.debug but after 1-2 minutes my USB keyboard and network >> > interface (remotely connected over SSH from my laptop) seemed to freeze >> and >> > I got disconnected. >> > >> > freebsd@generic:~ % uname -a >> > FreeBSD generic 14.0-CURRENT FreeBSD 14.0-CURRENT #0 >> > main-n253697-f1d450ddee6: Thu Mar 10 09:32:38 UTC 2022 >> > root@releng1.nyi.freebsd.org:/usr/obj/usr/src/arm64.aarch64/sys/GENERIC >> > arm64 >> > >> > root@generic:~ # sysctl -w hw.usb.dwc_otg.debug=1 >> > hw.usb.dwc_otg.debug: 0 -> 1 >> > >> > See attached initial debug info (no printing testing yet). >> > > > Hi Hans, > > >> >> Hi, >> >> Nothing obvious there. >> >> Maybe you need to add a conditional to DPRINTF's in the fast path, to >> only print for a certain device, to reduce the debug prints. >> >> How many USB devices are connected? >> > > Only two devices, one is my USB keyboard and the other one is my Epson > printer. > > >> >> One experiment you might try is to comment out: >> >> usbd_transfer_submit(xfer); >> >> in >> >> uhub_intr_callback() >> >> in >> >> /usr/src/sys/dev/usb/usb_hub.c >> >> It will save some USB resources, but also makes USB enumeration a bit >> slower. > > > This is noted, I'll try. > > >> Maybe the chip is out of USB HW endpoints and is only polling >> for RX data ... >> > > This is noted as well, we'll probably see later what's going on. I'll get > back to you once I recompile the kernel and perform the testing. > > Thanks, > Archimedes > Hi Hans, I did comment-out usbd_transfer_submit(xfer) and still the same behavior is experienced when hw.usb.dwc_otg.debug is enabled. The system seems to freeze as my USB keyboard is unresponsive and I cannot connect over the Ethernet NIC with SSH. I can't print and capture. freebsd@generic:/usr/src/sys/dev/usb % diff -Nur usb_hub.c.orig usb_hub.c --- usb_hub.c.orig 2022-03-14 22:37:03.162678000 +0800 +++ usb_hub.c 2022-03-14 22:38:23.832660000 +0800 @@ -202,7 +202,7 @@ case USB_ST_SETUP: usbd_xfer_set_frame_len(xfer, 0, usbd_xfer_max_len(xfer)); - usbd_transfer_submit(xfer); + /* usbd_transfer_submit(xfer); */ break; Let me know our next step. Thanks, Archimedes --000000000000c7449305da3aa2b0 Content-Type: text/html; charset="UTF-8" Content-Transfer-Encoding: quoted-printable <div dir=3D"ltr"><div dir=3D"ltr"><br></div><br><div class=3D"gmail_quote">= <div dir=3D"ltr" class=3D"gmail_attr">On Mon, Mar 14, 2022 at 6:12 PM Archi= medes Gaviola <<a href=3D"mailto:archimedes.gaviola@gmail.com">archimede= s.gaviola@gmail.com</a>> wrote:<br></div><blockquote class=3D"gmail_quot= e" style=3D"margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204)= ;padding-left:1ex"><div dir=3D"ltr"><div dir=3D"ltr"><br></div><br><div cla= ss=3D"gmail_quote"><div dir=3D"ltr" class=3D"gmail_attr">On Mon, Mar 14, 20= 22 at 5:31 PM Hans Petter Selasky <<a href=3D"mailto:hps@selasky.org" ta= rget=3D"_blank">hps@selasky.org</a>> wrote:<br></div><blockquote class= =3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-left:1px solid rg= b(204,204,204);padding-left:1ex">On 3/14/22 10:20, Archimedes Gaviola wrote= :<br> > On Sun, Mar 13, 2022 at 11:25 PM Archimedes Gaviola <<br> > <a href=3D"mailto:archimedes.gaviola@gmail.com" target=3D"_blank">arch= imedes.gaviola@gmail.com</a>> wrote:<br> > <br> >><br> >><br> >> On Sun, Mar 13, 2022 at 2:27 PM Archimedes Gaviola <<br> >> <a href=3D"mailto:archimedes.gaviola@gmail.com" target=3D"_blank">= archimedes.gaviola@gmail.com</a>> wrote:<br> >><br> >>><br> >>><br> >>> On Sun, Mar 13, 2022 at 12:38 AM Hans Petter Selasky <<a hr= ef=3D"mailto:hps@selasky.org" target=3D"_blank">hps@selasky.org</a>><br> >>> wrote:<br> >>><br> >>>> On 3/12/22 16:31, Archimedes Gaviola wrote:<br> >>>>>><br> >>>>>> IOERROR usually means an electrical error. The RPI= 3B will use a<br> >>>>>> transaction translator for the FULL speed traffic,= which is driven by<br> >>>>>> software.<br> >>>>><br> >>>><br> >>>> Hi Archimedes,<br> >>>><br> >>>>> Hi Hans,<br> >>>>><br> >>>>> I'm curious about the transaction translator you&#= 39;ve mentioned, any<br> >>>> idea why<br> >>>>> there's a need for translation and what is being t= ranslated?<br> >>>><br> >>>> When the High Speed USB HUB was invented, there was a need= to support<br> >>>> FULL and LOW speed USB transactions. Because FULL and LOW = speed<br> >>>> transactions are slow and take up much bandwidth, a transa= ctions<br> >>>> translator was invented which translates between High Spee= d USB and<br> >>>> FULL/LOW speed USB. That means the RPi 3B consists of a si= ngle USB HIGH<br> >>>> speed port, followed by a USB HUB. These transactions are = not visible in<br> >>>> usbdump .<br> >>>><br> >>>>=C2=A0 =C2=A0>Does this only<br> >>>>> happen when RPi 3B (acting as host controller) is tran= smitting data to<br> >>>> the<br> >>>>> Epson printer? Are translation events visible in the u= sbdump? In this<br> >>>> case<br> >>>>> there's a way to possibly track what's going o= n and how to identify any<br> >>>>> info that is being translated?<br> >>>><br> >>>> By turning on the HC debugging, you can possibly track via= debug<br> >>>> messages what is going on. Maybe it is a timing issue, tha= t the SW is<br> >>>> too slow serving the micro transactions.<br> >>>><br> >>>> Any idea also if translation is being<br> >>>>> performed in RPi 4B?<br> >>>><br> >>>> The later RPi's do the transaction translator bits in = HW or via the XHCI<br> >>>> block.<br> >>>><br> >>>> As I have conducted several printing cases with my<br> >>>>> Epson printer without any issues with either of the 4 = ports.<br> >>>>><br> >>>>> Sorry for these questions as I am catching-up to the U= SB technology<br> >>>>> internal workings under the hood.<br> >>>><br> >>>> You are welcome!<br> >>>><br> >>><br> >>><br> >>> Thank you so much Hans for answering my questions, really appr= eciate it!<br> >>> I have a much better understanding now.<br> >>><br> >>> I came from testing with 13.0-RELEASE having the same RPi 3B h= ardware and<br> >>> setup and it's very stable. I haven't encountered any = IOERROR and<br> >>> incomplete printed outputs that were encountered with 14.0-CUR= RENT<br> >>> (February 24, 2022). By the way, I found in the repository her= e<br> >>> <a href=3D"https://download.freebsd.org/snapshots/arm64/aarch6= 4/ISO-IMAGES/14.0/" rel=3D"noreferrer" target=3D"_blank">https://download.f= reebsd.org/snapshots/arm64/aarch64/ISO-IMAGES/14.0/</a><br> >>> that there were two latest snapshots released dated March 3 an= d March 10,<br> >>> respectively. I need to take printing tests first, especially = the latest to<br> >>> check if it also manifests before I go back to the Feb. 24 sna= pshot and do<br> >>> a thorough test with debugging. I'll provide updates for a= ny observations.<br> >>><br> >>> Thanks,<br> >>> Archimedes<br> >>><br> >><br> >><br> >> Hi Hans,<br> >><br> >> Initial testing conducted with the latest 14.0-CURRENT (March 10, = 2022<br> >> snapshot) seems to be stable. Another test will be performed tomor= row.<br> >><br> >> Thanks,<br> >> Archimedes<br> >><br> > <br> > Hi Hans,<br> > <br> > I still encountered the issue this morning with 14.0-CURRENT (March 10= ,<br> > 2022 snapshot). I just noticed when I left my RPi 3B idle for a long p= eriod<br> > of time (2-3 hours and more) and started printing, then the issue pops= -up.<br> > Another concern I encountered was when I started enabling<br> > hw.usb.dwc_otg.debug but after 1-2 minutes my USB keyboard and network= <br> > interface (remotely connected over SSH from my laptop) seemed to freez= e and<br> > I got disconnected.<br> > <br> > freebsd@generic:~ % uname -a<br> > FreeBSD generic 14.0-CURRENT FreeBSD 14.0-CURRENT #0<br> > main-n253697-f1d450ddee6: Thu Mar 10 09:32:38 UTC 2022<br> > root@releng1.nyi.freebsd.org:/usr/obj/usr/src/arm64.aarch64/sys/GENERI= C<br> > arm64<br> > <br> > root@generic:~ # sysctl -w hw.usb.dwc_otg.debug=3D1<br> > hw.usb.dwc_otg.debug: 0 -> 1<br> > <br> > See attached initial debug info (no printing testing yet).<br></blockq= uote><div><br></div><div><br></div><div>Hi Hans,</div><div>=C2=A0<br></div>= <blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-= left:1px solid rgb(204,204,204);padding-left:1ex"> <br> Hi,<br> <br> Nothing obvious there.<br> <br> Maybe you need to add a conditional to DPRINTF's in the fast path, to <= br> only print for a certain device, to reduce the debug prints.<br> <br> How many USB devices are connected?<br></blockquote><div><br></div><div>Onl= y two devices, one is my USB keyboard and the other one is my Epson printer= .</div><div>=C2=A0<br></div><blockquote class=3D"gmail_quote" style=3D"marg= in:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1e= x"> <br> One experiment you might try is to comment out:<br> <br> usbd_transfer_submit(xfer);<br> <br> in<br> <br> uhub_intr_callback()<br> <br> in<br> <br> /usr/src/sys/dev/usb/usb_hub.c<br> <br> It will save some USB resources, but also makes USB enumeration a bit <br> slower. </blockquote><div><br></div><div>This is noted, I'll try.<br></= div><div>=C2=A0<br></div><blockquote class=3D"gmail_quote" style=3D"margin:= 0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">= Maybe the chip is out of USB HW endpoints and is only polling <br> for RX data ...<br></blockquote><div><br></div><div>This is noted as well, = we'll probably see later what's going on. I'll get back to you = once I recompile the kernel and perform the testing.</div><div><br></div><d= iv>Thanks,</div><div>Archimedes<br></div></div></div></blockquote><div><br>= </div><div><br></div><div>Hi Hans,</div><div><br></div><div>I did comment-o= ut=20 usbd_transfer_submit(xfer) and still the same behavior is experienced when = hw.usb.dwc_otg.debug is enabled. The system seems to freeze as my USB keybo= ard is unresponsive and I cannot connect over the Ethernet NIC with SSH. I = can't print and capture.<br></div><div><br></div><div>freebsd@generic:/= usr/src/sys/dev/usb % diff -Nur usb_hub.c.orig usb_hub.c<br>--- usb_hub.c.o= rig =C2=A0 =C2=A0 =C2=A02022-03-14 22:37:03.162678000 +0800<br>+++ usb_hub.= c =C2=A0 2022-03-14 22:38:23.832660000 +0800<br>@@ -202,7 +202,7 @@<br><br>= =C2=A0 =C2=A0 =C2=A0 =C2=A0 case USB_ST_SETUP:<br>=C2=A0 =C2=A0 =C2=A0 =C2= =A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 usbd_xfer_set_frame_len(xfer, 0, usbd_xfer_= max_len(xfer));<br>- =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 usbd_= transfer_submit(xfer);<br>+ =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2= =A0 /* usbd_transfer_submit(xfer); */<br>=C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0= =C2=A0 =C2=A0 =C2=A0 break;</div><div><br></div><div>Let me know our next = step.</div><div><br></div><div>Thanks,</div><div>Archimedes<br></div><div>= =C2=A0</div></div></div> --000000000000c7449305da3aa2b0--
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CAJFbk7G_SVcoH4pak4SvNp0EYsbU_i4LiNZ6HyEDrmxAPvviUg>