| Message ID | 66ae38dc-1c0c-5a76-be23-fb87db90b327@free.fr (mailing list archive) |
|---|---|
| State | Accepted, archived |
| Commit | 915347f67d41857a514bed77053b212f3696e8a3 |
| Headers | show |
| Series | [v5] PCI: qcom: Use default config space read function | expand |
Bjorn, I think this patch is good enough to land now. Regards. On 25/03/2019 16:42, Marc Gonzalez wrote: > Move the device class fudge to a proper fixup function, and remove > qcom_pcie_rd_own_conf() which has become useless. > > NB: dw_pcie_setup_rc() already did the right thing, but it's broken > on older qcom chips, such as 8064. > > Signed-off-by: Marc Gonzalez <marc.w.gonzalez@free.fr> > --- > Changes from v4 to v5: Apply fixup to all qcom chips, the same way it was before > (thus the code remains functionally equivalent) > Drop Srinivas' Tested-by tag because of the change > --- > drivers/pci/controller/dwc/pcie-qcom.c | 23 ++++++----------------- > 1 file changed, 6 insertions(+), 17 deletions(-) > > diff --git a/drivers/pci/controller/dwc/pcie-qcom.c b/drivers/pci/controller/dwc/pcie-qcom.c > index a7f703556790..0ed235d560e3 100644 > --- a/drivers/pci/controller/dwc/pcie-qcom.c > +++ b/drivers/pci/controller/dwc/pcie-qcom.c > @@ -1129,25 +1129,8 @@ static int qcom_pcie_host_init(struct pcie_port *pp) > return ret; > } > > -static int qcom_pcie_rd_own_conf(struct pcie_port *pp, int where, int size, > - u32 *val) > -{ > - struct dw_pcie *pci = to_dw_pcie_from_pp(pp); > - > - /* the device class is not reported correctly from the register */ > - if (where == PCI_CLASS_REVISION && size == 4) { > - *val = readl(pci->dbi_base + PCI_CLASS_REVISION); > - *val &= 0xff; /* keep revision id */ > - *val |= PCI_CLASS_BRIDGE_PCI << 16; > - return PCIBIOS_SUCCESSFUL; > - } > - > - return dw_pcie_read(pci->dbi_base + where, size, val); > -} > - > static const struct dw_pcie_host_ops qcom_pcie_dw_ops = { > .host_init = qcom_pcie_host_init, > - .rd_own_conf = qcom_pcie_rd_own_conf, > }; > > /* Qcom IP rev.: 2.1.0 Synopsys IP rev.: 4.01a */ > @@ -1309,6 +1292,12 @@ static const struct of_device_id qcom_pcie_match[] = { > { } > }; > > +static void qcom_fixup_class(struct pci_dev *dev) > +{ > + dev->class = PCI_CLASS_BRIDGE_PCI << 8; > +} > +DECLARE_PCI_FIXUP_EARLY(PCI_VENDOR_ID_QCOM, PCI_ANY_ID, qcom_fixup_class); > + > static struct platform_driver qcom_pcie_driver = { > .probe = qcom_pcie_probe, > .driver = {
Hi Marc, Thanks for the patch! On 3/25/19 5:42 PM, Marc Gonzalez wrote: > Move the device class fudge to a proper fixup function, and remove > qcom_pcie_rd_own_conf() which has become useless. > > NB: dw_pcie_setup_rc() already did the right thing, but it's broken > on older qcom chips, such as 8064. > > Signed-off-by: Marc Gonzalez <marc.w.gonzalez@free.fr> > --- > Changes from v4 to v5: Apply fixup to all qcom chips, the same way it was before > (thus the code remains functionally equivalent) > Drop Srinivas' Tested-by tag because of the change > --- > drivers/pci/controller/dwc/pcie-qcom.c | 23 ++++++----------------- > 1 file changed, 6 insertions(+), 17 deletions(-) Acked-by: Stanimir Varbanov <svarbanov@mm-sol.com>
On Mon, Mar 25, 2019 at 04:42:55PM +0100, Marc Gonzalez wrote: > Move the device class fudge to a proper fixup function, and remove > qcom_pcie_rd_own_conf() which has become useless. > > NB: dw_pcie_setup_rc() already did the right thing, but it's broken > on older qcom chips, such as 8064. > > Signed-off-by: Marc Gonzalez <marc.w.gonzalez@free.fr> > --- > Changes from v4 to v5: Apply fixup to all qcom chips, the same way it was before > (thus the code remains functionally equivalent) > Drop Srinivas' Tested-by tag because of the change > --- > drivers/pci/controller/dwc/pcie-qcom.c | 23 ++++++----------------- > 1 file changed, 6 insertions(+), 17 deletions(-) Applied to pci/dwc for v5.2, thanks. Lorenzo > diff --git a/drivers/pci/controller/dwc/pcie-qcom.c b/drivers/pci/controller/dwc/pcie-qcom.c > index a7f703556790..0ed235d560e3 100644 > --- a/drivers/pci/controller/dwc/pcie-qcom.c > +++ b/drivers/pci/controller/dwc/pcie-qcom.c > @@ -1129,25 +1129,8 @@ static int qcom_pcie_host_init(struct pcie_port *pp) > return ret; > } > > -static int qcom_pcie_rd_own_conf(struct pcie_port *pp, int where, int size, > - u32 *val) > -{ > - struct dw_pcie *pci = to_dw_pcie_from_pp(pp); > - > - /* the device class is not reported correctly from the register */ > - if (where == PCI_CLASS_REVISION && size == 4) { > - *val = readl(pci->dbi_base + PCI_CLASS_REVISION); > - *val &= 0xff; /* keep revision id */ > - *val |= PCI_CLASS_BRIDGE_PCI << 16; > - return PCIBIOS_SUCCESSFUL; > - } > - > - return dw_pcie_read(pci->dbi_base + where, size, val); > -} > - > static const struct dw_pcie_host_ops qcom_pcie_dw_ops = { > .host_init = qcom_pcie_host_init, > - .rd_own_conf = qcom_pcie_rd_own_conf, > }; > > /* Qcom IP rev.: 2.1.0 Synopsys IP rev.: 4.01a */ > @@ -1309,6 +1292,12 @@ static const struct of_device_id qcom_pcie_match[] = { > { } > }; > > +static void qcom_fixup_class(struct pci_dev *dev) > +{ > + dev->class = PCI_CLASS_BRIDGE_PCI << 8; > +} > +DECLARE_PCI_FIXUP_EARLY(PCI_VENDOR_ID_QCOM, PCI_ANY_ID, qcom_fixup_class); > + > static struct platform_driver qcom_pcie_driver = { > .probe = qcom_pcie_probe, > .driver = { > -- > 2.17.1
On 29/03/2019 17:38, Lorenzo Pieralisi wrote: > On Mon, Mar 25, 2019 at 04:42:55PM +0100, Marc Gonzalez wrote: > >> Move the device class fudge to a proper fixup function, and remove >> qcom_pcie_rd_own_conf() which has become useless. >> >> NB: dw_pcie_setup_rc() already did the right thing, but it's broken >> on older qcom chips, such as 8064. >> >> Signed-off-by: Marc Gonzalez <marc.w.gonzalez@free.fr> >> --- >> Changes from v4 to v5: Apply fixup to all qcom chips, the same way it was before >> (thus the code remains functionally equivalent) >> Drop Srinivas' Tested-by tag because of the change >> --- >> drivers/pci/controller/dwc/pcie-qcom.c | 23 ++++++----------------- >> 1 file changed, 6 insertions(+), 17 deletions(-) > > Applied to pci/dwc for v5.2, thanks. Thanks! Do you push to linux-next? (The patch is not in next-20190401) Regards.
On Mon, Apr 01, 2019 at 12:39:33PM +0200, Marc Gonzalez wrote: > On 29/03/2019 17:38, Lorenzo Pieralisi wrote: > > > On Mon, Mar 25, 2019 at 04:42:55PM +0100, Marc Gonzalez wrote: > > > >> Move the device class fudge to a proper fixup function, and remove > >> qcom_pcie_rd_own_conf() which has become useless. > >> > >> NB: dw_pcie_setup_rc() already did the right thing, but it's broken > >> on older qcom chips, such as 8064. > >> > >> Signed-off-by: Marc Gonzalez <marc.w.gonzalez@free.fr> > >> --- > >> Changes from v4 to v5: Apply fixup to all qcom chips, the same way it was before > >> (thus the code remains functionally equivalent) > >> Drop Srinivas' Tested-by tag because of the change > >> --- > >> drivers/pci/controller/dwc/pcie-qcom.c | 23 ++++++----------------- > >> 1 file changed, 6 insertions(+), 17 deletions(-) > > > > Applied to pci/dwc for v5.2, thanks. > > Thanks! > > Do you push to linux-next? (The patch is not in next-20190401) Bjorn will push it into linux-next in due course. Thanks, Lorenzo
diff --git a/drivers/pci/controller/dwc/pcie-qcom.c b/drivers/pci/controller/dwc/pcie-qcom.c index a7f703556790..0ed235d560e3 100644 --- a/drivers/pci/controller/dwc/pcie-qcom.c +++ b/drivers/pci/controller/dwc/pcie-qcom.c @@ -1129,25 +1129,8 @@ static int qcom_pcie_host_init(struct pcie_port *pp) return ret; } -static int qcom_pcie_rd_own_conf(struct pcie_port *pp, int where, int size, - u32 *val) -{ - struct dw_pcie *pci = to_dw_pcie_from_pp(pp); - - /* the device class is not reported correctly from the register */ - if (where == PCI_CLASS_REVISION && size == 4) { - *val = readl(pci->dbi_base + PCI_CLASS_REVISION); - *val &= 0xff; /* keep revision id */ - *val |= PCI_CLASS_BRIDGE_PCI << 16; - return PCIBIOS_SUCCESSFUL; - } - - return dw_pcie_read(pci->dbi_base + where, size, val); -} - static const struct dw_pcie_host_ops qcom_pcie_dw_ops = { .host_init = qcom_pcie_host_init, - .rd_own_conf = qcom_pcie_rd_own_conf, }; /* Qcom IP rev.: 2.1.0 Synopsys IP rev.: 4.01a */ @@ -1309,6 +1292,12 @@ static const struct of_device_id qcom_pcie_match[] = { { } }; +static void qcom_fixup_class(struct pci_dev *dev) +{ + dev->class = PCI_CLASS_BRIDGE_PCI << 8; +} +DECLARE_PCI_FIXUP_EARLY(PCI_VENDOR_ID_QCOM, PCI_ANY_ID, qcom_fixup_class); + static struct platform_driver qcom_pcie_driver = { .probe = qcom_pcie_probe, .driver = {
Move the device class fudge to a proper fixup function, and remove qcom_pcie_rd_own_conf() which has become useless. NB: dw_pcie_setup_rc() already did the right thing, but it's broken on older qcom chips, such as 8064. Signed-off-by: Marc Gonzalez <marc.w.gonzalez@free.fr> --- Changes from v4 to v5: Apply fixup to all qcom chips, the same way it was before (thus the code remains functionally equivalent) Drop Srinivas' Tested-by tag because of the change --- drivers/pci/controller/dwc/pcie-qcom.c | 23 ++++++----------------- 1 file changed, 6 insertions(+), 17 deletions(-)