PowerNV PCIe hotplug support?

[Timothy Pearson]

Hrmm, potentially related, I'm getting a kernel oops when I try to hot unplug the entire PCIe switch from the PHB4:

[ 3771.106478] Kernel attempted to read user page (10) - exploit attempt? (uid: 0)
[ 3771.106514] BUG: Kernel NULL pointer dereference on read at 0x00000010
[ 3771.106531] Faulting instruction address: 0xc000000000da0f44
[ 3771.106558] Oops: Kernel access of bad area, sig: 11 [#1]
[ 3771.106573] LE PAGE_SIZE=64K MMU=Radix SMP NR_CPUS=2048 NUMA PowerNV
[ 3771.106610] Modules linked in: pnv_php i2c_dev binfmt_misc xts ecb ctr nvme nvme_core t10_pi cbc crc64_rocksoft_generic ofpart drm_shmem_helper drm_kms_helper crc64_rocksoft vmx_crypto joydev evdev crc64 i2c_algo_bit gf128mul at24 ipmi_powernv powernv_flash crct10dif_vpmsum ipmi_devintf opal_prd regmap_i2c mtd ipmi_msghandler drm loop fuse drm_panel_orientation_quirks dm_mod ip_tables x_tables autofs4 nfsv3 nfs_acl nfs lockd grace sunrpc fscache netfs ib_mthca ib_ipoib ib_umad rdma_ucm rdma_cm iw_cm ib_cm configfs mlx4_ib ib_uverbs ib_core hid_generic usbhid hid xhci_pci xhci_hcd ixgbe mdio_devres of_mdio crc32c_vpmsum tg3 fixed_phy fwnode_mdio xfrm_algo mdio mlx4_core usbcore libphy usb_common
[ 3771.106879] CPU: 4 PID: 1173 Comm: bash Tainted: G        W          6.6.8 #1
[ 3771.106908] Hardware name: T2P9D01 REV 1.00 POWER9 0x4e1203 opal:skiboot-9858186 PowerNV
[ 3771.106955] NIP:  c000000000da0f44 LR: c000000000da0f38 CTR: c0000000008b58e0
[ 3771.106983] REGS: c0000000125a37a0 TRAP: 0300   Tainted: G        W           (6.6.8)
[ 3771.107011] MSR:  9000000000009033 <SF,HV,EE,ME,IR,DR,RI,LE>  CR: 24424222  XER: 00000006
[ 3771.107054] CFAR: c000000000d9d96c DAR: 0000000000000010 DSISR: 40000000 IRQMASK: 0
[ 3771.107054] GPR00: c000000000da0f38 c0000000125a3a40 c0000000010fe800 0000000000000000
[ 3771.107054] GPR04: 0000000000000040 0000000000000040 0000000000000009 0000000000000000
[ 3771.107054] GPR08: c000201ffcba1b00 0000000000000000 c00000000ebbcf00 c008000015ca3700
[ 3771.107054] GPR12: c0000000008b58e0 c000001ffffdba00 0000000000000000 0000000114bfa548
[ 3771.107054] GPR16: 0000000000000000 0000000000000000 0000000000000000 0000000114b23e40
[ 3771.107054] GPR20: 000000012468c760 0000000114bb1a58 00007fffca6d7214 c000201fff7933c0
[ 3771.107054] GPR24: 0000000000000001 0000000000000000 c000201fff793b20 c000000011004460
[ 3771.107054] GPR28: c000000011004440 c000000013b3b000 c000000013b3b0c0 0000000000000010
[ 3771.107349] NIP [c000000000da0f44] mutex_lock+0x34/0x90
[ 3771.107379] LR [c000000000da0f38] mutex_lock+0x28/0x90
[ 3771.107407] Call Trace:
[ 3771.107425] [c0000000125a3a40] [c000000000da0f38] mutex_lock+0x28/0x90 (unreliable)
[ 3771.107459] [c0000000125a3a70] [c000000000202658] msi_lock_descs+0x28/0x40
[ 3771.107490] [c0000000125a3a90] [c0000000008b5948] pci_disable_msi+0x68/0xb0
[ 3771.107520] [c0000000125a3ac0] [c008000015ca0708] pnv_php_disable_irq+0x140/0x160 [pnv_php]
[ 3771.107556] [c0000000125a3b10] [c008000015ca0a28] pnv_php_free_slot+0x50/0x90 [pnv_php]
[ 3771.107588] [c0000000125a3b40] [c008000015ca1fb0] pnv_php_unregister+0x248/0x290 [pnv_php]
[ 3771.107629] [c0000000125a3bb0] [c008000015ca207c] pnv_php_disable_slot+0x84/0xe0 [pnv_php]
[ 3771.107671] [c0000000125a3bf0] [c0000000008cf8d0] power_write_file+0xa0/0x180
[ 3771.107701] [c0000000125a3c70] [c0000000008c4b60] pci_slot_attr_store+0x40/0x60
[ 3771.107733] [c0000000125a3c90] [c000000000614ef4] sysfs_kf_write+0x64/0x80
[ 3771.107763] [c0000000125a3cb0] [c000000000613a18] kernfs_fop_write_iter+0x1b8/0x2a0
[ 3771.107794] [c0000000125a3d00] [c000000000521070] vfs_write+0x350/0x4b0
[ 3771.107831] [c0000000125a3dc0] [c0000000005214d4] ksys_write+0x84/0x140
[ 3771.107861] [c0000000125a3e10] [c000000000030d38] system_call_exception+0x168/0x3a0
[ 3771.107902] [c0000000125a3e50] [c00000000000c270] system_call_vectored_common+0xf0/0x280
[ 3771.107959] --- interrupt: 3000 at 0x7fffb834ad38
[ 3771.107993] NIP:  00007fffb834ad38 LR: 0000000000000000 CTR: 0000000000000000
[ 3771.108030] REGS: c0000000125a3e80 TRAP: 3000   Tainted: G        W           (6.6.8)
[ 3771.108048] MSR:  900000000280f033 <SF,HV,VEC,VSX,EE,PR,FP,ME,IR,DR,RI,LE>  CR: 48422488  XER: 00000000
[ 3771.108107] IRQMASK: 0
[ 3771.108107] GPR00: 0000000000000004 00007fffca6d6ff0 00007fffb8466f00 0000000000000001
[ 3771.108107] GPR04: 0000000124681100 0000000000000002 0000000000000010 0000000000000030
[ 3771.108107] GPR08: 0000000000000000 0000000000000000 0000000000000000 0000000000000000
[ 3771.108107] GPR12: 0000000000000000 00007fffb857b0c0 0000000000000000 0000000114bfa548
[ 3771.108107] GPR16: 0000000000000000 0000000000000000 0000000000000000 0000000114b23e40
[ 3771.108107] GPR20: 000000012468c760 0000000114bb1a58 00007fffca6d7214 00007fffca6d7210
[ 3771.108107] GPR24: 0000000000000001 0000000114c006f0 0000000124681100 0000000000000002
[ 3771.108107] GPR28: 0000000000000002 00007fffb8461980 0000000124681100 0000000000000002
[ 3771.108380] NIP [00007fffb834ad38] 0x7fffb834ad38
[ 3771.108404] LR [0000000000000000] 0x0
[ 3771.108425] --- interrupt: 3000
[ 3771.108437] Code: 3842d8f0 7c0802a6 60000000 7c0802a6 fbe1fff8 7c7f1b78 f8010010 f821ffd1 4bffc9f5 60000000 39200000 e94d0908 <7d00f8a8> 7c284800 40c20010 7d40f9ad
[ 3771.108510] ---[ end trace 0000000000000000 ]---

----- Original Message -----
> From: "Timothy Pearson" <tpearson at raptorengineeringinc.com>
> To: "linuxppc-dev" <linuxppc-dev at lists.ozlabs.org>
> Sent: Wednesday, December 27, 2023 10:15:24 PM
> Subject: PowerNV PCIe hotplug support?

> I've been evaluating some new options for our POWER9-based hardware in the NVMe
> space, and would like some clarification on the current status of PCIe hotplug
> for the PowerNV platforms.
> 
> From what I understand, the pnv_php driver provides the basic hotplug
> functionality on PowerNV.  What I'm not clear on is to what extent this is
> intended to flow downstream to attached PCIe switches.
> 
> I have a test setup here that consists of a PMC 8533 switch and several
> downstream NVMe drives, with the switch attached directly to the PHB4 root
> port.  After loading the pnv_php module, I can disconnect the downstream NVMe
> devices by either using echo 0 on /sys/bus/pcu/slots/Snnnnnnn/power, or by
> doing a physical surprise unplug, however nothing I try can induce a newly
> plugged device to train and be detected on the bus.  Even trying a echo 0 and
> subsequent echo 1 to /sys/bus/pcu/slots/Snnnnnnn/power only results in the
> device going offline, there seems to be no way to bring the device back online
> short of a reboot.
> 
> Hotplug of other devices connected directly to the PHB4 appears to work properly
> (I can online and offline via the power node); the issue seems to be restricted
> to downstream devices connected to the (theoretically hotplug capable) PMC 8533
> switch.
> 
> Is this the intended behavior for downstream (non-IBM) PCIe ports?  Raptor can
> provide resources to assist in a fix if needed, but I would like to understand
> if this is a bug or an unimplemented feature first, and if the latter what the
> main issues are likely to be in implementation.
> 
> Thank you!