2010/12/26 Michael Karcher <Michael.Karcher@fu-berlin.de>
Am Montag, den 06.12.2010, 18:01 +0100 schrieb Idwer Vollering:
> This corrects a djgpp build error, seen with r1232 and later.
>
> Signed-off-by: Idwer Vollering <vidwer@gmail.com>
Not acked...

> -int rpci_write_long(struct pci_dev *dev, int reg, uint32_t data)
> +int rpci_write_long(struct pci_dev *dev, int reg, u32 data)

> pcidev.c:210: error: conflicting types for 'rpci_write_long'
> programmer.h:226: error: previous declaration of 'rpci_write_long' was
> here

I'd rather patch the header file to use uint32_t/uint16_t/uint8_t
instead of u8/u16/u32, as the former types are used throughout flashrom.

Patch v2.

Signed-off-by: Idwer Vollering <vidwer@gmail.com>

Index: programmer.h
===================================================================
--- programmer.h        (revision 1247)
+++ programmer.h        (working copy)
@@ -221,9 +221,9 @@
 /* rpci_write_* are reversible writes. The original PCI config space register
  * contents will be restored on shutdown.
  */
-int rpci_write_byte(struct pci_dev *dev, int reg, u8 data);
-int rpci_write_word(struct pci_dev *dev, int reg, u16 data);
-int rpci_write_long(struct pci_dev *dev, int reg, u32 data);
+int rpci_write_byte(struct pci_dev *dev, int reg, uint8_t data);
+int rpci_write_word(struct pci_dev *dev, int reg, uint16_t data);
+int rpci_write_long(struct pci_dev *dev, int reg, uint32_t data);
 #endif

 /* print.c */
 
Is there any reason to not use them on rpci_write_*?

No, using the same type in a header as well as in the file that includes that header makes sense.
Compile- and runtime tested using make (linux and djgpp) and gmake respectively.


Regards,
 Michael Karcher