Use the raid controller LSI MegaRAID SAS 9271-8i. #uname -a FreeBSD ... 10.3-STABLE FreeBSD 10.3-STABLE #0: Wed Jul 13 16:45:24 MSK 2016 ... amd64 # mfiutil show adapter mfi0 Adapter: Product Name: LSI MegaRAID SAS 9271-8i Serial Number: SV42934928 Firmware: 23.34.0-0005 RAID Levels: JBOD, RAID0, RAID1, RAID5, RAID6, RAID10, RAID50 Battery Backup: present NVRAM: 32K Onboard Memory: 1024M Minimum Stripe: 8K Maximum Stripe: 1M After upgrading to 11.0-Stable, revision: 302669, raid controller became invisible.
Have you tried the mrsas driver? I'm CCing one of the driver maintainers for more input. kadesai hasn't registered in Bugzilla yet (he maintains one of the other drivers). pciconf -lv and /var/run/dmesg.boot output would help too.
Created attachment 172494 [details] pciconf.log
(In reply to Sergey Renkas from comment #2) Is this from when you're booted into 10.3-STABLE ?
(In reply to Ngie Cooper from comment #3) Yes! In the 11th version of the mfid device not shown.
(In reply to Sergey Renkas from comment #4) Ok. pciconf -lv and /var/run/dmesg.boot are more valuable from 11.x than 10.3-STABLE, but having both of those items from 10.3-STABLE would be helpful as well for comparing things.
(In reply to Ngie Cooper from comment #5) It will take some time to run a live-cd 11 version.
Created attachment 172495 [details] picconf.log 11 version
Created attachment 172496 [details] dmesg.boot 11 version
(In reply to Sergey Renkas from comment #6) Files already added.
(In reply to Sergey Renkas from comment #7) The device isn't present here, i.e. there's nothing for mfi(4) to attach to according to the PCI bus. This output at boot looks interesting... unknown: I/O range not supported That exact message doesn't pop up on my machine. It just says, atrtc0: Warning: Couldn't map I/O. Could you please boot 11.0 with boot -v and attach the output of /var/run/dmesg.boot?
Created attachment 172498 [details] dmesg.boot 11 version boot -v
(In reply to Ngie Cooper from comment #10) is ready
Hi I am Kashyap..manage <mrsas> driver for LSI/Avago MR controller. Let me know if I have to look at this. From this discussion and attached dmesg/pci logs, I can see that device is not even detected at PCI layer, so driver is not involved for Device not getting detected. Also could not able to find anything odd from logs... May be others in this BZ can point out ..
(In reply to Sergey Renkas from comment #11) Ok. Something's not working from the ACPI/PCI end of things. Is your BIOS up to date? Are you using the traditional BIOS boot or a UEFI boot? What's your motherboard model?
(In reply to Ngie Cooper from comment #14) Supermicro X7DBU. "traditional BIOS" 2.1a 12/20/08 FreeBSD 8-9-10 they all worked ... 11 non funziona! )
Please capture 'pciconf -lcBbev' output from 11. The PLX PCI-express switch is HotPlug capable and it seems that the bridge is not being enumerated. The pciconf output will tell me if the PLX chip is claiming that there is nothing plugged in. (For future reference, pciconf -lv doesn't have the really useful bits, the useful bits are generally in -c and to a lesser extent -b, -B, and -e). You can just run this against the relevant bridges btw, e.g.: 'pciconf -lcbBev pcib11'
Created attachment 172545 [details] pciconf_lcbBev_pcib11
(In reply to John Baldwin from comment #16) done...waiting for a miracle.)
Comment on attachment 172545 [details] pciconf_lcbBev_pcib11 Interesting, this PCI-PCI bridge has many more of the "optional" hot plug features. In this case it claims that there is a mechanical latch (MRL) that holds the card in that is open. We refuse to attach to the bus if the MRL is open (and in general we try to power down the slot assuming that a user has opened the latch to remove the card). Can you see any sort of latch near the slot the card is in?
(In reply to John Baldwin from comment #19) There is no latch. Can send pictures to your E-Mail.
Is the LSI adapter in a physical slot or is it integrated onto the motherboard? If you can send me pictures that would be nice, perhaps one with no card in the slot at all if possible?
(In reply to John Baldwin from comment #21) Is the LSI adapter in a physical slot... sent pictures to your E-Mail.(In reply to John Baldwin from comment #21) Is the LSI adapter in a physical slot... sent picture to your E-Mail.
So this add-on card doesn't appear to honor the spec. It should be wiring the sensor closed or some such if it isn't implemented. OTOH, it seems like the Linux hotplug code (by my reading) doesn't actually care if the sensor is open. As a test, can you try patching pcib_hotplug_inserted() in sys/dev/pci/pci_pci.c to comment out the last check? (The lines under the 'If the MRL is disengaged...' comment)
(In reply to John Baldwin from comment #23) The card is detected. But at mount time the error was detected: interrupt strom detected on "irq259" throtting interrupt source ... and so on to infinity.
Grr, ok. Let's try to narrow down why it breaks. I will probably add a tunable to disable hotplug entirely, though that will require a manual tunable to be set for this box unfortunately. Can you drop your earlier patch and instead hack pcib_probe_hotplug() to always be empty (in particular, to never set PCIB_HOTPLUG)?
(In reply to John Baldwin from comment #25) Hacking pcib_probe_hotplug() was successful. pcib_probe_hotplug(struct pcib_softc *sc) { } The disk was successfully mounted.
Created attachment 173071 [details] pci_hp_tunable.patch Please try the attached patch. You will need to set 'hw.pci.enable_pcie_hp=0' in /boot/loader.conf for now. I still want to try to figure out what write to the slot control register is breaking your device, but this tunable is something I can merge to 11.0 at least.
(In reply to John Baldwin from comment #27) Thanks, the patch works.
A commit references this bug: Author: jhb Date: Fri Jul 29 17:54:21 UTC 2016 New revision: 303497 URL: https://svnweb.freebsd.org/changeset/base/303497 Log: Add a loader tunable (hw.pci.enable_pcie_hp) to disable PCI-e HotPlug. Some systems and/or devices (such as riser cards) do not include a non-compliant implementation of PCI-e HotPlug that can result in devices not being attached (e.g. the HotPlug code might assume that a card is being unplugged and will power the slot off and detach it). This tunable can be set to 0 to disable support for PCI-e HotPlug ignoring the incorrect HotPlug state on these slots. PR: 211081 Reported by: Sergey Renkas <serg_ic@mail.ru> (SuperMicro X7 riser card) Reported by: Jeffrey E Pieper <jeffrey.e.pieper@intel.com> (Intel X520 adapter) MFC after: 1 week Relnotes: yes Changes: head/sys/dev/pci/pci_pci.c
A commit references this bug: Author: jhb Date: Mon Aug 1 22:19:23 UTC 2016 New revision: 303645 URL: https://svnweb.freebsd.org/changeset/base/303645 Log: Disable PCI hotplug support for slots with power controllers. After further review of the spec, I do not think the current HotPlug code handles slots with power controllers correctly. In particular, the power state of the slot is to be inferred from other events, not from examining the state of the power control bit in SLOT_CTL. For now, disable PCI hotplug support on such slots. PR: 211081 Tested by: Jeffrey E Pieper <jeffrey.e.pieper@intel.com> MFC after: 3 days Changes: head/sys/dev/pci/pci_pci.c
A commit references this bug: Author: jhb Date: Fri Aug 5 18:41:51 UTC 2016 New revision: 303781 URL: https://svnweb.freebsd.org/changeset/base/303781 Log: MFC 303497,303559,303645: Disable PCI-e hotplug on bridges with power controllers. 303497: Add a loader tunable (hw.pci.enable_pcie_hp) to disable PCI-e HotPlug. Some systems and/or devices (such as riser cards) do not include a non-compliant implementation of PCI-e HotPlug that can result in devices not being attached (e.g. the HotPlug code might assume that a card is being unplugged and will power the slot off and detach it). This tunable can be set to 0 to disable support for PCI-e HotPlug ignoring the incorrect HotPlug state on these slots. 303559: Try to declare _hw_pci for all sysctl cases needed after r303497. 303645: Disable PCI hotplug support for slots with power controllers. After further review of the spec, I do not think the current HotPlug code handles slots with power controllers correctly. In particular, the power state of the slot is to be inferred from other events, not from examining the state of the power control bit in SLOT_CTL. For now, disable PCI hotplug support on such slots. PR: 211081 Approved by: re (gjb) Changes: _U stable/11/ stable/11/sys/dev/pci/pci_pci.c