[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20180312140316.GB194000@bhelgaas-glaptop.roam.corp.google.com>
Date: Mon, 12 Mar 2018 09:03:16 -0500
From: Bjorn Helgaas <helgaas@...nel.org>
To: Christoph Hellwig <hch@...radead.org>
Cc: linux-pci@...r.kernel.org, linux-kernel@...r.kernel.org,
"Rafael J. Wysocki" <rjw@...ysocki.net>, linux-pm@...r.kernel.org,
Keith Busch <keith.busch@...el.com>,
Sinan Kaya <okaya@...eaurora.org>,
Lukas Wunner <lukas@...ner.de>,
Frederick Lawler <fred@...dlawl.com>
Subject: Re: [PATCH v2 08/13] PCI/portdrv: Simplify PCIe feature permission
checking
On Mon, Mar 12, 2018 at 01:04:02AM -0700, Christoph Hellwig wrote:
> > + * We assume we can manage these PCIe features. Some systems may
> > + * reserve these for use by the platform itself, e.g., an ACPI BIOS
> > + * may implement its own AER handling and use _OSC to prevent the
> > + * OS from interfering.
> > + */
> > + bridge->use_aer = 1;
> > + bridge->use_hotplug = 1;
> > + bridge->use_pme = 1;
>
> If we start out with enabled maybe these should be disable_foo flags
> instead?
I went back and forth on that. "disable_foo" is nice because the
default value is correct (zero means enabled). But then you end up
with things like:
if (pcie_ports_native || !host->disable_hotplug)
where the "!host->disable_hotplug" is a double negative, and I have a
really hard time reading that.
> Also please use bool (or a bitfield bool) for true/false values.
I'm a little ambivalent about bool in structs because of things like
this:
https://lkml.kernel.org/r/CA+55aFxnePDimkVKVtv3gNmRGcwc8KQ5mHYvUxY8sAQg6yvVYg@mail.gmail.com
Bjorn
Powered by blists - more mailing lists