Skip to contentSkip to conversation

powerpc/fsl-pci: Unify pci/pcie initialization code

DISCUSSION Daniel Walker 2 participants 4 messages Nov 5, 2015

lore
Share & tools
lore
#0 lore raw

Hi,

905e75c46dba5f3061049277e4eb7110beedba43:arch/powerpc/platforms/85xx/mpc85xx_ads.c

This commit seems to cause problems with PCI devices doing DMA, just with mpc85xx ads. If you do the "fsl_add_bridge()" after you setup the ppc_md.pci_exclude_device to mpc85xx_exclude_device, then the early_read_config_word() calls in arch/powerpc/sysdev/fsl_pci.c:setup_pci_cmd() are like no-ops.

This issue was found by David Beazley (on the CC), he can respond to more detailed questions on it. The fix he came up with was to remove the "pci_exclude_device".

Daniel

#1 lore raw
Quoted 11 lines · Daniel Walker

Hi,

905e75c46dba5f3061049277e4eb7110beedba43:arch/powerpc/platforms/85xx/mpc85xx_ads.c

This commit seems to cause problems with PCI devices doing DMA, just with mpc85xx ads. If you do the "fsl_add_bridge()" after you setup the ppc_md.pci_exclude_device to mpc85xx_exclude_device, then the early_read_config_word() calls in arch/powerpc/sysdev/fsl_pci.c:setup_pci_cmd() are like no-ops.

Quoted 5 lines

This issue was found by David Beazley (on the CC), he can respond to more detailed questions on it. The fix he came up with was to remove the "pci_exclude_device".

Daniel

OK, please send a patch.

-Scott

#2 lore raw
Quoted 12 lines · Scott Wood
Quoted 11 lines · Daniel Walker

Hi,

905e75c46dba5f3061049277e4eb7110beedba43:arch/powerpc/platforms/85xx/mpc85xx_ads.c

This commit seems to cause problems with PCI devices doing DMA, just with mpc85xx ads. If you do the "fsl_add_bridge()" after you setup the ppc_md.pci_exclude_device to mpc85xx_exclude_device, then the early_read_config_word() calls in arch/powerpc/sysdev/fsl_pci.c:setup_pci_cmd() are like no-ops.

Quoted 7 lines
Quoted 5 lines

This issue was found by David Beazley (on the CC), he can respond to more detailed questions on it. The fix he came up with was to remove the "pci_exclude_device".

Daniel

OK, please send a patch.

I was hoping Jia Hongtao could address why it wasn't address in the original patch. Some of the other boards also have an exclude function, and I wonder if those work with this change.

Daniel

#3 lore raw
Quoted 18 lines · Daniel Walker
Quoted 17 lines · Scott Wood
Quoted 16 lines · Daniel Walker

Hi,

905e75c46dba5f3061049277e4eb7110beedba43:arch/powerpc/platforms/85xx/mpc85xx_ads.c

This commit seems to cause problems with PCI devices doing DMA, just with mpc85xx ads. If you do the "fsl_add_bridge()" after you setup the ppc_md.pci_exclude_device to mpc85xx_exclude_device, then the early_read_config_word() calls in arch/powerpc/sysdev/fsl_pci.c:setup_pci_cmd() are like no-ops.

This issue was found by David Beazley (on the CC), he can respond to more detailed questions on it. The fix he came up with was to remove the "pci_exclude_device".

Quoted 7 lines
Quoted 3 lines
Quoted 1 line

Daniel

OK, please send a patch.

I was hoping Jia Hongtao could address why it wasn't address in the original patch. Some of the other boards also have an exclude function, and I wonder if those work with this change.

This is the only board I saw that excludes the host bridge. It wasn't addressed because it wasn't noticed.

-Scott