* pci_alloc_irq_vectors problem further investigation
@ 2022-12-12 10:35 Danilo Cezar Zanella
2022-12-19 13:30 ` Daniel Wagner
0 siblings, 1 reply; 2+ messages in thread
From: Danilo Cezar Zanella @ 2022-12-12 10:35 UTC (permalink / raw)
To: linux-rt-users
Hi,
There is a driver developed by EmbeddedTS (the board is Marvell
armada-385, dual core processor) that tries to allocate 32 MSI
interrupts, using pci_alloc_irq_vectors(..,32,32,...). When I ran the
Kernel (4.14 NON RT) provided by them, it worked fine, but when I
recompiled the kernel (4.19.249 using the RT_FULL options), it didn't,
returning -ENOSPC.
In RT_FULL Kernel:
- First fact, when I ran pci_msi_vec_count, it returned 32.
- Second, if I changed to 16, calling
pci_alloc_irq_vectors(..,16,16,...), it worked.
I traced this kernel and I found that:
pci_alloc_irq_vectors_affinity
__pci_enable_msi_range
msi_capability_init
pci_msi_setup_msi_irqs
msi_domain_alloc_irqs
__irq_domain_alloc_irqs
irq_domain_alloc_descs
Then I stopped here...
I searched on the Internet, but I couldn't find anything.
Has someone had a similar problem?
Is there something that I missed? Some configuration??
Thanks
--
Danilo C. Zanella
Radio Astronomy Group / IAG
University of Sao Paulo
Rua do Matão, 1226 - Cidade Universitária São Paulo - SP - Brasil - 05508-090
Tel.: +55 11 3091-2743
^ permalink raw reply [flat|nested] 2+ messages in thread
* Re: pci_alloc_irq_vectors problem further investigation
2022-12-12 10:35 pci_alloc_irq_vectors problem further investigation Danilo Cezar Zanella
@ 2022-12-19 13:30 ` Daniel Wagner
0 siblings, 0 replies; 2+ messages in thread
From: Daniel Wagner @ 2022-12-19 13:30 UTC (permalink / raw)
To: Danilo Cezar Zanella; +Cc: linux-rt-users
On Mon, Dec 12, 2022 at 07:35:14AM -0300, Danilo Cezar Zanella wrote:
> There is a driver developed by EmbeddedTS (the board is Marvell
> armada-385, dual core processor) that tries to allocate 32 MSI
> interrupts, using pci_alloc_irq_vectors(..,32,32,...). When I ran the
> Kernel (4.14 NON RT) provided by them, it worked fine, but when I
> recompiled the kernel (4.19.249 using the RT_FULL options), it didn't,
> returning -ENOSPC.
Does the kernel 'provided by them' carry additional patches for your
board?
> In RT_FULL Kernel:
> - First fact, when I ran pci_msi_vec_count, it returned 32.
> - Second, if I changed to 16, calling
> pci_alloc_irq_vectors(..,16,16,...), it worked.
> I traced this kernel and I found that:
> pci_alloc_irq_vectors_affinity
> __pci_enable_msi_range
> msi_capability_init
> pci_msi_setup_msi_irqs
> msi_domain_alloc_irqs
> __irq_domain_alloc_irqs
> irq_domain_alloc_descs
> Then I stopped here...
>
> I searched on the Internet, but I couldn't find anything.
> Has someone had a similar problem?
> Is there something that I missed? Some configuration??
Vendor kernels usually ship changes which are not ported to upstream and
hence upstream kernels might not work or work differently.
Daniel
^ permalink raw reply [flat|nested] 2+ messages in thread
end of thread, other threads:[~2022-12-19 13:30 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2022-12-12 10:35 pci_alloc_irq_vectors problem further investigation Danilo Cezar Zanella
2022-12-19 13:30 ` Daniel Wagner
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.