| Submitter | Rudolf Marek |
|---|---|
| Date | 2010-10-06 22:31:57 |
| Message ID | <4CACF8DD.6050509@assembler.cz> |
| Download | mbox | patch |
| Permalink | /patch/2059/ |
| State | Accepted |
| Headers | show |
Comments
On Thu, Oct 07, 2010 at 12:31:57AM +0200, Rudolf Marek wrote: > Hello, > > Following patch fixes the boot_switch_sata_ide logic. It can swap > primary / secondary IDE channel with SATA (in IDE mode). > > The bug was that setup was done in wrong device. > > Signed-off-by: Rudolf Marek <r.marek@assembler.cz> Yep, indeed. Acked-by: Uwe Hermann <uwe@hermann-uwe.de> > Index: sb700_ide.c > =================================================================== > --- sb700_ide.c (revision 5917) > +++ sb700_ide.c (working copy) > @@ -52,10 +53,13 @@ > > /* set ide as primary, if you want to boot from IDE, you'd better set it > * in $vendor/$mainboard/devicetree.cb */ > + > + > if (conf->boot_switch_sata_ide == 1) { > - byte = pci_read_config8(dev, 0xAD); > + struct device *sm_dev = dev_find_slot(0, PCI_DEVFN(0x14, 0)); I personally don't like the "struct device *sm_dev" to be defined here, please consider putting it at the beginning of the function, not in the middle (yeah, I know, it works, I just don't like it much). Uwe.
On Sun, Oct 10, 2010 at 12:05 PM, Uwe Hermann <uwe@hermann-uwe.de> wrote: *snip* >> =================================================================== >> --- sb700_ide.c (revision 5917) >> +++ sb700_ide.c (working copy) >> @@ -52,10 +53,13 @@ >> >> /* set ide as primary, if you want to boot from IDE, you'd better set it >> * in $vendor/$mainboard/devicetree.cb */ >> + >> + >> if (conf->boot_switch_sata_ide == 1) { >> - byte = pci_read_config8(dev, 0xAD); >> + struct device *sm_dev = dev_find_slot(0, PCI_DEVFN(0x14, 0)); > > I personally don't like the "struct device *sm_dev" to be defined here, > please consider putting it at the beginning of the function, not in the > middle (yeah, I know, it works, I just don't like it much). Doesn't declaring the struct there mean that it's only available within that block's scope? Is there any reason to declare the var outside that block if that's the only place where the var is used? Thanks, wt
Patch
Index: sb700_ide.c =================================================================== --- sb700_ide.c (revision 5917) +++ sb700_ide.c (working copy) @@ -52,10 +53,13 @@ /* set ide as primary, if you want to boot from IDE, you'd better set it * in $vendor/$mainboard/devicetree.cb */ + + if (conf->boot_switch_sata_ide == 1) { - byte = pci_read_config8(dev, 0xAD); + struct device *sm_dev = dev_find_slot(0, PCI_DEVFN(0x14, 0)); + byte = pci_read_config8(sm_dev, 0xAD); byte |= 1 << 4; - pci_write_config8(dev, 0xAD, byte); + pci_write_config8(sm_dev, 0xAD, byte); } #if CONFIG_PCI_ROM_RUN == 1
Hello, Following patch fixes the boot_switch_sata_ide logic. It can swap primary / secondary IDE channel with SATA (in IDE mode). The bug was that setup was done in wrong device. Signed-off-by: Rudolf Marek <r.marek@assembler.cz> Thanks, Rudolf