From mboxrd@z Thu Jan 1 00:00:00 1970 From: arnd@arndb.de (Arnd Bergmann) Date: Wed, 06 May 2015 17:33:33 +0200 Subject: [PATCH v3 3/5] PCI: st: Provide support for the sti PCIe controller In-Reply-To: References: <1428657168-12495-1-git-send-email-gabriel.fernandez@linaro.org> Message-ID: <12753608.WrdVVJIhrQ@wuerfel> To: linux-arm-kernel@lists.infradead.org List-Id: linux-arm-kernel.lists.infradead.org On Wednesday 06 May 2015 15:40:39 Gabriel Fernandez wrote: > On 6 May 2015 at 15:29, Bjorn Helgaas wrote: > > I don't really care which solution you end up with here. But please > > do take a look at how the other drivers solve the same problem. Using > > "device_initcall()" is unique in drivers/pci/host, and I don't believe > > the problem is unique. If several drivers have the same issue, they > > should solve it the same way. > > I agree with last reply of Arnd. > I will mark the probe function as __init and use > platform_driver_probe() to prevent the probe function from being > deferred. > And to prevent module unloading, i will use module_init only. > > Ok i will wait before sending a v4. I think in order to address Bjorn's concern, you should modify make sure that the keystone and imx6 drivers that call hook_fault_code() do it the same way. I think imx6 already does, but keystone does not, so please send a patch for it, explaining why it's needed. Conversely, I think the other dw-pcie based drivers should be changed not to use an __init annotation at all, and instead use module_platform_driver(). Arnd