[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <bec9b233-5069-c54a-4c11-9e6c33e7ef9a@siemens.com>
Date: Thu, 5 Jan 2017 08:03:13 +0100
From: Jan Kiszka <jan.kiszka@...mens.com>
To: Christoph Hellwig <hch@....de>
Cc: Andy Shevchenko <andy.shevchenko@...il.com>,
Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
Andy Shevchenko <andriy.shevchenko@...ux.intel.com>,
"linux-serial@...r.kernel.org" <linux-serial@...r.kernel.org>,
Linux Kernel Mailing List <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH] serial: 8250_lpss: Release Quark MSI vectors on exit
On 2017-01-05 06:25, Christoph Hellwig wrote:
> On Wed, Jan 04, 2017 at 11:46:58PM +0100, Jan Kiszka wrote:
>>> I NAKed already third patch related to PCI managed resources (couple
>>> of those regarding to pci_irq_* API)/
>>>
>>
>> Ah, there are resources that are managed without being allocated
>> explicitly that way. Hmm, not very intuitive. Are MSI / MSI-X vectors
>> the only such cases?
>
> MSI/MSI-X resources are not managed that way and an explicit call to
> pci_free_irq_vectors is required from the API standpoint. It might not
> actually free memory in many cases, but it still is a symmetric API.
>
Andy is referring to the following:
- pcim_enable_device registers pci_devres with the devres subsystem
- on device release, devres_release_all is invoked, and that calls
pcim_release
- the latter will invoke pci_disable_msi and pci_disable_msix if any of
them is enabled
The user will still call the same pci_msi_enable, pci_alloc_irq_vectors
etc., but they are now "magically" managed with pcim_enable_device being
used. Same for pci_request_region.
Jan
--
Siemens AG, Corporate Technology, CT RDA ITP SES-DE
Corporate Competence Center Embedded Linux
Powered by blists - more mailing lists