New: Kernel 4.19.135: pci_resource_start, pci_resource_len return wrong value
[+cc linux-pci]
Quoted 11 lines · On Wed, Oct 05, 2022 at 01:00:58PM +0000, [email protected] wrote:
https://bugzilla.kernel.org/show_bug.cgi?id=216555
Summary: Kernel 4.19.135:pci_resource_start,pci_resource_len
return wrong valueI used 4.19.135(x64) + rootfs (buildroot) to boot an x86 SBC. This SBC is connected via PCIe to a customized PCIe IO card.
In the kernel module, I used
pci_resource_start,pci_resource_lento find start, length of 3 BARs.
It worked fine.
Quoted 6 lines
Then I used the same kernel, kernel module on another x86 SBC. In this SBC,
pci_resource_start,pci_resource_lenreturned wrong data. I booted this SBC with Centos 8.2 (4.18) and also with vanilla 4.9.20. It worked fine.Is it possible that 4.19.135 has a BUG ?
It's unlikely that pci_resource_start() and pci_resource_len() are broken in 4.19.135, but I guess anything is possible.
Can you give any more details about why it seems broken? Dmesg logs of a boot that fails and one that works, for instance?
Bjorn
Quoted 17 lines · Bjorn Helgaas
Quoted 16 lines · On Wed, Oct 05, 2022 at 01:00:58PM +0000, [email protected] wrote:
https://bugzilla.kernel.org/show_bug.cgi?id=216555
Summary: Kernel 4.19.135:pci_resource_start,pci_resource_len
return wrong valueI used 4.19.135(x64) + rootfs (buildroot) to boot an x86 SBC. This SBC is connected via PCIe to a customized PCIe IO card.
In the kernel module, I used
pci_resource_start,pci_resource_lento find start, length of 3 BARs.
It worked fine.Then I used the same kernel, kernel module on another x86 SBC. In this SBC,
pci_resource_start,pci_resource_lenreturned wrong data. I booted this SBC with Centos 8.2 (4.18) and also with vanilla 4.9.20. It worked fine.
Quoted 7 lines
Quoted 1 line
Is it possible that 4.19.135 has a BUG ?
It's unlikely that
pci_resource_start()andpci_resource_len()are broken in 4.19.135, but I guess anything is possible.Can you give any more details about why it seems broken? Dmesg logs of a boot that fails and one that works, for instance?
I see that you attached a dmesg log and some more details to the bugzilla, thank you!
I don't see "AdtIoInit" in the kernel tree, and I can't match the [1172:0001] with anything. Can you also attach the driver source to the bugzilla?
Bjorn