11-Stable no longer works with my USB controllers. My old kernel, r314236, works fine, but my kernel of April 6 (r316552) does not see any USB devices. Any recent changes to the USB software that might explain this? I will be trying to track it down by bisecting, but this will be slow die to lack of time. (Moving in two days.) I see the following from the boot: ehci0: <EHCI (generic) USB 2.0 controller> mem 0xf252a000-0xf252a3ff irq 16 at device 26.0 on pci0 usbus0: EHCI version 1.0 usbus0 on ehci0 [...] ehci1: <EHCI (generic) USB 2.0 controller> mem 0xf2529000-0xf25293ff irq 23 at device 29.0 on pci0 sbus1: EHCI version 1.0 usbus1 on ehci1 [...] usbus0: 480Mbps High Speed USB v2.0 usbus1: 480Mbps High Speed USB v2.0 ugen1.1: <Intel EHCI root HUB> at usbus1 uhub0: <Intel EHCI root HUB, class 9/0, rev 2.00/1.00, addr 1> on usbus1 ugen0.1: <Intel EHCI root HUB> at usbus0 uhub1: <Intel EHCI root HUB, class 9/0, rev 2.00/1.00, addr 1> on usbus0 uhub0: 3 ports with 3 removable, self powered uhub1: 3 ports with 3 removable, self powered usbus0: port reset timeout usbus1: port reset timeout uhub_reattach_port: port 1 reset failed, error=USB_ERR_TIMEOUT uhub_reattach_port: device problem (USB_ERR_TIMEOUT), disabling port 1 uhub_reattach_port: port 1 reset failed, error=USB_ERR_TIMEOUT uhub_reattach_port: device problem (USB_ERR_TIMEOUT), disabling port 1 usbconfig -d ugen1.1 reset produced: Apr 9 09:15:11 rogue kernel: uhub1: at usbus0, port 1, addr 1 (disconnected) Apr 9 09:15:11 rogue kernel: uhub1: Apr 9 09:15:11 rogue kernel: <Intel EHCI root HUB, class 9/0, rev 2.00/1.00, addr 1> on usbus0 Apr 9 09:15:12 rogue kernel: uhub1: 3 ports with 3 removable, self powered From pcoconf -lv: ehci0@pci0:0:26:0: class=0x0c0320 card=0x21cf17aa chip=0x1c2d8086 rev=0x04 hdr=0x00 vendor = 'Intel Corporation' device = '6 Series/C200 Series Chipset Family USB Enhanced Host Controller' class = serial bus subclass = USB I have an almost GENERIC kernel with only: nooptions SCHED_ULE # ULE scheduler options SCHED_4BSD # 4BSD scheduler options IEEE80211_DEBUG src.conf contains: ogue# cat /etc/src.conf PORTS_MODULES=emulators/virtualbox-ose-kmod sysutils/lsof WITHOUT_ATM=true WITHOUT_FLOPPY=true WITHOUT_GPIB=YES WITHOUT_I4B==YES WITHOUT_IPFILTER=true WITHOUT_LPR=true WITHOUT_NCP=YES WITHOUT_PF=YES WITHOUT_PPP=true WITHOUT_PROFILE=YES WITHOUT_SENDMAIL=true
The issue is the update to llvm-4.0. With r316422, USB works fine. llvm-4.0 was MFCed in r316423 and USB is not working with that kernel. The commit did not touch and USB specific code. Should dim and/or emaste be involved? I can't think pulling the update is reasonable, but losing USB support i pretty serious, too.
Hi, You might want to do "objdump -Dx /boot/kernel/kernel" and compare the output from the two different compilers. --HPS
(In reply to Hans Petter Selasky from comment #2) The dumps are 441 MB and 438 MB and the kernels are significantly different. just the size differs by about 171 MB. They may be downloaded from http://ykoberman.dlinkddns.com/FreeBSD. ATM I am in the process of moving. It was my intent to make a backup of my system prior to packing it that caused me to use a USB drive and that is how I discovered the issue. It also means that my server will be down for a couple of days starting (probably) in about 5 or 6 hours, though I may leave it up overnight as it takes little time to pack up and travels in a separate vehicle from most stuff (where the backup disks sit). Would it be useful to have the kernels there?
localhost(11.0-S)[5] uname -a FreeBSD localhost 11.0-STABLE FreeBSD 11.0-STABLE #315 r317083M/317085:1100512: Tue Apr 18 05:32:22 PDT 2017 root@localhost:/common/S1/obj/usr/src/sys/CANARY amd64 localhost(11.0-S)[6] tail -0F /var/log/messages Apr 18 19:57:39 localhost kernel: ugen0.2: <JMicron USB Mass Storage Device> at usbus0 Apr 18 19:57:39 localhost kernel: umass0 on uhub0 Apr 18 19:57:39 localhost kernel: umass0: <MSC Bulk-Only Transfer> on usbus0 Apr 18 19:57:39 localhost kernel: umass0: SCSI over Bulk-Only; quirks = 0x8100 Apr 18 19:57:39 localhost kernel: umass0:5:0: Attached to scbus5 Apr 18 19:57:39 localhost kernel: da0 at umass-sim0 bus 0 scbus5 target 0 lun 0 Apr 18 19:57:39 localhost kernel: da0: <M4-CT512 M4SSD2 0103> Fixed Direct Access SPC-4 SCSI device Apr 18 19:57:39 localhost kernel: da0: Serial Number DB9876543211161 Apr 18 19:57:39 localhost kernel: da0: 400.000MB/s transfers Apr 18 19:57:39 localhost kernel: da0: 488386MB (1000215216 512 byte sectors) Apr 18 19:57:39 localhost kernel: da0: quirks=0x2<NO_6_BYTE> [That was the handiest USB device I happened to have....]
Hi, Maybe you can repeat the dump for usb,ehci,uhci,ohci,xhci .ko only. Do you feel certain the compiler was properly built? --HPS
Dumps of the usb, ehci, uhci, xhci and ohci modules are all now available at http://ykoberman.dlinkddns.com/FreeBSD/ and are labeled for 316422 (LLVM3.9) and 316423 (LLVM4.0). Note that this system has no xhci capability. David Wolfskill has success with a newer Lynx Point based system. This may be limited to the Cougar Point chipset. Are there good places I might insert printf statement in the code to further narrow this down? I plan to build a 316423 kernel without USB support so modules may be loaded to make testing and debugging easier. I realize there there is a possibility that this kernel may work. The compiler built without any fatal error with a standard "make -j6 buildworld".
This is really feeling like a race. I built a kernel without any USB devices. I boot to single user and, when I load the USB device drivers, everything works. I can plug in a thumb drive and it works normally. If I boot and load the USB drivers at the "ok" loader prompt and then boot, the USB fails. I can only guess that, with the 4.0 compiled kernel, ehci or usb load before something is ready and it fails to properly initialize. Not sure where to go from here, but I will update my stable system to the latest and see how things work.
Was any resolution to this issue found? I'm having problems installing onto a Dell R630 with 11-RELEASE/STABLE (kernel can't see USB boot media), 12-CURRENT (bootloader freezes) and 10.4-RELEASE (panic) as stock mini-memstick downloads. For reference Xen 7.2 is working on the box, so I know the hardware is good and the keyboard works at the loader. I'm was about to upgrade the BIOS before troubling the ML and saw this.
(In reply to leeb from comment #8) I should close this ticket. It was clearly a race and it vanished after a system update shortly after the discussion of this ticket. It was simply a failure to connect to the USB hub at boot time and even slight adjustments that affected timing eliminated the issue. I'd suggest that you post a message about these problems you are seeing to the usb@freebsd.org list. Include details on exactly what you are seeing. hte more detailed, the better. I do know that the current (12.0) boot is in a state of flux and has been a bit unstable, but 10 and 11 should be fine. Sorry that I can't be of more help.