Fix dereferencing the free'd oct_dev pointer.

[gammadj19]

Hello,
I'm a member of Montis vRAN NIC team in Samsung.
I found a dereferencing free'd pointer bug which makes segmentation fault error during rmmod oct_ep_phc as below:
Please review and pull this fix.
Thanks,

# rmmod oct_ep_phc
[ 878.354856] Octeon EP PHC 0001:01:00.1: OCT_PHC[0]: Stopping octeon device
[ 878.361814] Octeon EP PHC 0001:01:00.1: OCT_PHC[0]: Freeing PCI mapped regions for Bar0
[ 878.369816] Octeon EP PHC 0001:01:00.1: OCT_PHC[0]: Freeing PCI mapped regions for Bar1
[ 878.377809] Octeon EP PHC 0001:01:00.1: OCT_PHC[0]: Freeing PCI mapped regions for Bar2
[ 878.385803] Octeon EP PHC 0001:01:00.1: OCT_PHC[0]: BAR unmapped.
[ 878.391900] Unable to handle kernel paging request at virtual address ffff8000080d7078
[ 878.399833] Mem abort info:
[ 878.402624] ESR = 0x96000007
[ 878.405671] EC = 0x25: DABT (current EL), IL = 32 bits
[ 878.410977] SET = 0, FnV = 0
[ 878.414024] EA = 0, S1PTW = 0
[ 878.417159] FSC = 0x07: level 3 translation fault
[ 878.422032] Data abort info:
[ 878.424905] ISV = 0, ISS = 0x00000007
[ 878.428733] CM = 0, WnR = 0
[ 878.431693] swapper pgtable: 4k pages, 48-bit VAs, pgdp=00000000f3995000
[ 878.438385] [ffff8000080d7078] pgd=10000083fffff003, p4d=10000083fffff003, pud=10000083ffffe003, pmd=10000083ffffd003, pte=0000000000000000
[ 878.450899] Internal error: Oops: 96000007 [#1] SMP
[ 878.455763] Modules linked in: oct_ep_phc(E-) nls_iso8859_1 aes_ce_blk aes_ce_cipher crct10dif_ce ghash_ce sha2_ce sha256_arm64 sha1_ce plx_dma efi_pstore arm_cmn arm_spe_pmu sbsa_gwdt sch_fq_codel vfio_pci vfio_pci_core vfio_virqfd vfio_iommu_type1 vfio ip_tables x_tables autofs4 r8169 xhci_pci ahci xhci_pci_renesas realtek [last unloaded: oct_ep_phc]
[ 878.486949] CPU: 0 PID: 578 Comm: rmmod Tainted: G E 5.18.4+ #1
[ 878.494157] pstate: 80400009 (Nzcv daif +PAN -UAO -TCO -DIT -SSBS BTYPE=--)
[ 878.501104] pc : octeon_ep_phc_remove+0xfc/0x128 [oct_ep_phc]
[ 878.506838] lr : octeon_ep_phc_remove+0xfc/0x128 [oct_ep_phc]
[ 878.512570] sp : ffff80000b4b3ae0
[ 878.515871] x29: ffff80000b4b3ae0 x28: ffff00800add9080 x27: 0000000000000000
[ 878.522992] x26: ffff80000a6cac60 x25: ffff008000cd2150 x24: ffff80000a6cac78
[ 878.530114] x23: ffff800001273208 x22: ffff8000080d7208 x21: 0000000000000000
[ 878.537235] x20: ffff8000012730c0 x19: ffff8000080d7000 x18: 0000000000000000
[ 878.544357] x17: 0000000000000000 x16: 0000000000000000 x15: 0000000000000000
[ 878.551477] x14: ffff80000a398468 x13: ffff80000a3978a0 x12: 0000000000000000
[ 878.558599] x11: 0000000000000004 x10: 0000000000000c40 x9 : 0000000000000000
[ 878.565720] x8 : ffff00800add9d20 x7 : ffff0080008e5a10 x6 : ffff80000a35e000
[ 878.572841] x5 : 1d90630814364341 x4 : fffffc0200240f20 x3 : 000000008020001b
[ 878.579962] x2 : 0000000000000000 x1 : 0000000000000000 x0 : ffff800001273a90
[ 878.587084] Call trace:
[ 878.589517] octeon_ep_phc_remove+0xfc/0x128 [oct_ep_phc]
[ 878.594902] pci_device_remove+0x48/0xe0
[ 878.598814] device_remove+0x54/0x90
[ 878.602377] device_release_driver_internal+0x204/0x2b4
[ 878.607590] driver_detach+0x5c/0xec
[ 878.611151] bus_remove_driver+0x78/0x130
[ 878.615148] driver_unregister+0x3c/0x6c
[ 878.619057] pci_unregister_driver+0x30/0xa4
[ 878.623313] phc_exit+0x1c/0x708 [oct_ep_phc]
[ 878.627656] __arm64_sys_delete_module+0x18c/0x2b0
[ 878.632435] invoke_syscall+0x78/0x100
[ 878.636171] el0_svc_common.constprop.0+0x58/0x190
[ 878.640949] do_el0_svc+0x30/0x90
[ 878.644251] el0_svc+0x4c/0x1b4
[ 878.647380] el0t_64_sync_handler+0xe8/0xf0
[ 878.651550] el0t_64_sync+0x1a0/0x1a4
[ 878.655200] Code: aa1303e0 97fffd1a aa1303e0 97fffc98 (f9403e60)
[ 878.661279] ---[ end trace 0000000000000000 ]---
Segmentation fault

[jiam-marvell]

Hi,
Our Host PCIe Driver engineer review your patch.
First, Really appreciate for pointing out this. It is valid bug.
Second, our engineer think it doesn't seem fulfilling. We will fix it and solution will be sync to this repo/branch alone with our future Monthly SDK update.
Lastly, Contact our FAE to create support ticket, you will get update from FAE when the patch is ready to release. If you don't know which FAE you should contact, let me know and leave your email address, I will send contact info through email