On Wed, May 27, 2026 at 11:38:14PM +0530, Aditya Gupta wrote:
> For opencapi phb direct slots, the .pdev for php_slots will be NULL
> 
> Various sections of the code in pnv_php can do a null dereference and
> crash the kernel.
> 
> Originally, the issue was hit during boot:
> 
>     [    1.568588] PowerPC PowerNV PCI Hotplug Driver version: 0.1
>     [    1.569722] BUG: Kernel NULL pointer dereference at 0x00000074
>     [    1.569811] Faulting instruction address: 0xc000000000b75fd0
>     [    1.569890] Oops: Kernel access of bad area, sig: 11 [#1]
>     [    1.569963] LE PAGE_SIZE=64K MMU=Hash  SMP NR_CPUS=2048 NUMA PowerNV
>     ...
>     [    1.571492] NIP [c000000000b75fd0] pnv_php_get_adapter_state+0x60/0x154
>     [    1.571604] LR [c000000000b75fbc] pnv_php_get_adapter_state+0x4c/0x154
>     [    1.571690] Call Trace:
>     [    1.571725] [c000c0000688f990] [c000000000b75fbc] 
> pnv_php_get_adapter_state+0x4c/0x154 (unreliable)
>     [    1.571783] [c000c0000688fa20] [c000000000b78bd0] 
> pnv_php_enable+0x94/0x378
>     [    1.571951] [c000c0000688fac0] [c000000000b7912c] 
> pnv_php_register_one.isra.0+0x11c/0x1e0

Drop timestamps since they don't add useful information.

Indent quoted material by two spaces to reduce wrapping.

Run "git log --oneline drivers/pci/hotplug/pnv_php.c" and "git log
--oneline drivers/pci/hotplug/" and match subject line style.

> This occurs for hotplug slots on root buses where bus->self == NULL,
> such as OpenCAPI PHB direct slots. An added debug print (not part of
> this patch) confirmed it was opencapi:

Style "OpenCAPI" and "PHB" consistently in commit log and subject.

>     [    1.617227] pnv_php: slot 'OPENCAPI-0009' has NULL pdev (bus 0009:00, 
> parent=NO (root bus))
>     [    1.617308] pnv_php: slot 'OPENCAPI-0009' 
> dn->full_name='pciex@603a000000000', 
> compatible='ibm,power10-pau-opencapi-pciex'
> 
> This only required null check in 'pnv_php_get_adapter_state', which
> caused the kernel to boot.
> 
> Even with 'pnv_php_get_adapter_state' null check, there are more
> possible null dereferences pointed by sashiko, including cases where
> userspace crashes the kernel, such as:
> 
>     $ cat /sys/bus/pci/slots/*/attention
>     ...
>     [  557.036295] Kernel attempted to read user page (6e) - exploit attempt? 
> (uid: 0)
>     [  557.036354] BUG: Kernel NULL pointer dereference on read at 0x0000006e
>     [  557.036383] Faulting instruction address: 0xc000000000a83334
>     [  557.036413] Oops: Kernel access of bad area, sig: 11 [#1]
>     [  557.036449] LE PAGE_SIZE=64K MMU=Hash  SMP NR_CPUS=2048 NUMA PowerNV
>     ...
>     [  557.037749] [c000000046707a20] [c000000046707b90] 0xc000000046707b90 
> (unreliable)
>     [  557.037795] [c000000046707a70] [0000000000000001] 0x1
>     [  557.037850] [c000000046707ab0] [c000000000acb00c] 
> attention_read_file+0x54/0xa8
>     [  557.037910] [c000000046707b30] [c000000000abfbfc] 
> pci_slot_attr_show+0x3c/0x58
>     [  557.037977] [c000000046707b50] [c0000000008181ec] 
> sysfs_kf_seq_show+0xd4/0x204
>     [  557.038022] [c000000046707be0] [c000000000815004] 
> kernfs_seq_show+0x44/0x58
> 
> Add null checks to prevent the null dereferences.
> 
> Cc: [email protected]
> Fixes: 80f9fc236279 ("PCI: pnv_php: Work around switches with broken presence 
> detection")
> Signed-off-by: Aditya Gupta <[email protected]>

Reply via email to