2006-10-06 02:17:49

by Olof Johansson

[permalink] [raw]
Subject: [PATCH] powerpc: fix fsl_soc build breaks

Hrm, there's no way this ever built at time of merge. There's a missing } and
the wrong type on phy_irq.

Also, another const for get_property().


CC arch/powerpc/sysdev/fsl_soc.o
arch/powerpc/sysdev/fsl_soc.c: In function 'fs_enet_of_init':
arch/powerpc/sysdev/fsl_soc.c:625: error: assignment of read-only variable 'phy_irq'
arch/powerpc/sysdev/fsl_soc.c:625: warning: assignment makes integer from pointer without a cast
arch/powerpc/sysdev/fsl_soc.c:661: warning: assignment discards qualifiers from pointer target type
arch/powerpc/sysdev/fsl_soc.c:684: error: subscripted value is neither array nor pointer
arch/powerpc/sysdev/fsl_soc.c:687: error: subscripted value is neither array nor pointer
arch/powerpc/sysdev/fsl_soc.c:722: warning: ISO C90 forbids mixed declarations and code
arch/powerpc/sysdev/fsl_soc.c:728: error: invalid storage class for function 'cpm_uart_of_init'
arch/powerpc/sysdev/fsl_soc.c:798: error: initializer element is not constant
arch/powerpc/sysdev/fsl_soc.c:798: error: expected declaration or statement at end of input
make[1]: *** [arch/powerpc/sysdev/fsl_soc.o] Error 1


Signed-off-by: Olof Johansson <[email protected]>


---

There are more issues with this file. Whitespace, if () {}; and other
things. I'm just fixing the build breaks.

These were all introduced by patches fed upstream via git trees instead
of list posts, as far as I can tell. Maybe posting patches is a better
idea, more eyes on the code.


Index: linux-2.6/arch/powerpc/sysdev/fsl_soc.c
===================================================================
--- linux-2.6.orig/arch/powerpc/sysdev/fsl_soc.c
+++ linux-2.6/arch/powerpc/sysdev/fsl_soc.c
@@ -567,7 +567,7 @@ static int __init fs_enet_of_init(void)
struct resource r[4];
struct device_node *phy, *mdio;
struct fs_platform_info fs_enet_data;
- const unsigned int *id, *phy_addr, phy_irq;
+ const unsigned int *id, *phy_addr, *phy_irq;
const void *mac_addr;
const phandle *ph;
const char *model;
@@ -641,7 +641,7 @@ static int __init fs_enet_of_init(void)

if (strstr(model, "FCC")) {
int fcc_index = *id - 1;
- unsigned char* mdio_bb_prop;
+ const unsigned char *mdio_bb_prop;

fs_enet_data.dpram_offset = (u32)cpm_dpram_addr(0);
fs_enet_data.rx_ring = 32;
@@ -708,8 +708,9 @@ static int __init fs_enet_of_init(void)
ret = platform_device_add_data(fs_enet_dev, &fs_enet_data,
sizeof(struct
fs_platform_info));
- if (ret)
- goto unreg;
+ if (ret)
+ goto unreg;
+ }
}
return 0;


2006-10-06 02:36:06

by Vitaly Bordug

[permalink] [raw]
Subject: Re: [PATCH] powerpc: fix fsl_soc build breaks

On Thu, 5 Oct 2006 21:16:48 -0500
Olof Johansson wrote:

> Hrm, there's no way this ever built at time of merge. There's a
> missing } and the wrong type on phy_irq.
>
> Also, another const for get_property().
>
>
> CC arch/powerpc/sysdev/fsl_soc.o
> arch/powerpc/sysdev/fsl_soc.c: In function 'fs_enet_of_init':
> arch/powerpc/sysdev/fsl_soc.c:625: error: assignment of read-only
> variable 'phy_irq' arch/powerpc/sysdev/fsl_soc.c:625: warning:
> assignment makes integer from pointer without a cast
> arch/powerpc/sysdev/fsl_soc.c:661: warning: assignment discards
> qualifiers from pointer target type
> arch/powerpc/sysdev/fsl_soc.c:684: error: subscripted value is
> neither array nor pointer arch/powerpc/sysdev/fsl_soc.c:687: error:
> subscripted value is neither array nor pointer
> arch/powerpc/sysdev/fsl_soc.c:722: warning: ISO C90 forbids mixed
> declarations and code arch/powerpc/sysdev/fsl_soc.c:728: error:
> invalid storage class for function 'cpm_uart_of_init'
> arch/powerpc/sysdev/fsl_soc.c:798: error: initializer element is not
> constant arch/powerpc/sysdev/fsl_soc.c:798: error: expected
> declaration or statement at end of input make[1]: ***
> [arch/powerpc/sysdev/fsl_soc.o] Error 1
>
>
> Signed-off-by: Olof Johansson <[email protected]>
>
>
> ---
>
> There are more issues with this file. Whitespace, if () {}; and other
> things. I'm just fixing the build breaks.
>
> These were all introduced by patches fed upstream via git trees
> instead of list posts, as far as I can tell. Maybe posting patches is
> a better idea, more eyes on the code.
>

All those were submitted in form of patches prior, and located issues addressed. I must mistype something with git-push,
so it didn't fed in very latest changes. Thanks for pointing it out and sorry for confusion.

--
Sincerely, Vitaly


Attachments:
signature.asc (189.00 B)