This patch fixes an issue in config space read accessors: we should fill-in the value even if we fail (e.g. skipping devices), otherwise CONFIG_PCI_SCAN_SHOW reports bogus values during boot up.
Signed-off-by: Anton Vorontsov avorontsov@ru.mvista.com --- cpu/mpc83xx/pcie.c | 9 +++++++-- 1 files changed, 7 insertions(+), 2 deletions(-)
diff --git a/cpu/mpc83xx/pcie.c b/cpu/mpc83xx/pcie.c index 1a3b2be..77f8906 100644 --- a/cpu/mpc83xx/pcie.c +++ b/cpu/mpc83xx/pcie.c @@ -60,6 +60,9 @@ static int mpc83xx_pcie_remap_cfg(struct pci_controller *hose, pci_dev_t dev) #define cfg_write(val, addr, type, op) \ do { op((type *)(addr), (val)); } while (0)
+#define cfg_read_err(val) do { *val = -1; } while (0) +#define cfg_write_err(val) do { } while (0) + #define PCIE_OP(rw, size, type, op) \ static int pcie_##rw##_config_##size(struct pci_controller *hose, \ pci_dev_t dev, int offset, \ @@ -68,8 +71,10 @@ static int pcie_##rw##_config_##size(struct pci_controller *hose, \ int ret; \ \ ret = mpc83xx_pcie_remap_cfg(hose, dev); \ - if (ret) \ - return ret; \ + if (ret) { \ + cfg_##rw##_err(val); \ + return ret; \ + } \ cfg_##rw(val, (void *)hose->cfg_addr + offset, type, op); \ return 0; \ }