On 09/04/25 2:16 am, Thomas Gleixner wrote:
On Tue, Apr 08 2025 at 18:20, Bert Karwatzki wrote:
Am Dienstag, dem 08.04.2025 um 17:29 +0200 schrieb Thomas Gleixner:
Can you please decode the lines please via:
scripts/faddr2line vmlinux msi_domain_first_desc+0x4/0x30
scripts/faddr2line vmlinux msix_setup_interrupts+0x23b/0x280
I had to recompile with CONFIG_DEBUG_INFO=Y, and reran the test, the calltrace
is identical.
$ scripts/faddr2line vmlinux msi_domain_first_desc+0x4/0x30
msi_domain_first_desc+0x4/0x30:
msi_domain_first_desc at kernel/irq/msi.c:400
So it seems msi_domain_first_desc() is called with dev = NULL.
This issue is seen on IBM Power9 server also. Bisection is pointing to
7b025f3f85ed4283d5a414371bb2ffd38d19033f as first bad commit.
I applied this patch on top of 20250409 and it fixes the issue.
Yup
$ scripts/faddr2line vmlinux msix_setup_interrupts+0x23b/0x280
msix_setup_interrupts+0x23b/0x280:
msix_update_entries at drivers/pci/msi/msi.c:647 (discriminator 1)
Aaarg. The patch below should fix that.
Thanks,
tglx
---
diff --git a/drivers/pci/msi/msi.c b/drivers/pci/msi/msi.c
index 4027abcafe7a..77cc27e45b66 100644
--- a/drivers/pci/msi/msi.c
+++ b/drivers/pci/msi/msi.c
@@ -680,8 +680,8 @@ static int __msix_setup_interrupts(struct pci_dev *__dev,
struct msix_entry *ent
if (ret)
return ret;
- retain_ptr(dev);
msix_update_entries(dev, entries);
+ retain_ptr(dev);
return 0;
}
Reported-by: Venkat Rao Bagalkote <venka...@linux.ibm.com>
Tested-by: Venkat Rao Bagalkote <venka...@linux.ibm.com>
Regards,
Venkat.
