Hello.
Cyril Chemparathy wrote:
This patch allows for a more flexible ioremap() interception based on iotable
contents.
SoCs that need ioremap() to fill up davinci_soc_info may directly call
davinci_io_init() and use ioremap() as needed. Such SoCs must not fill
io_desc and io_desc_num entries in struct davinci_soc_info.
Signed-off-by: Cyril Chemparathy <[email protected]>
Tested-by: Sandeep Paulraj <[email protected]>
---
v4: removed __davinci_ioremap() and related infrastructure
v4: split davinci_io_init() out of davinci_common_init()
[...]
diff --git a/arch/arm/mach-davinci/io.c b/arch/arm/mach-davinci/io.c
index a1c0b6b..81862ec 100644
--- a/arch/arm/mach-davinci/io.c
+++ b/arch/arm/mach-davinci/io.c
[...]
@@ -36,3 +48,30 @@ void davinci_iounmap(volatile void __iomem *addr)
__iounmap(addr);
}
EXPORT_SYMBOL(davinci_iounmap);
+
+void __init davinci_io_init(struct map_desc *desc, unsigned long desc_num)
+{
+ if (!desc || !desc_num)
+ return;
+
+ /* silently refuse to init twice ...*/
+ if (davinci_io_desc) {
+ /* ... unless things change */
+ BUG_ON(davinci_io_desc != desc ||
+ davinci_io_desc_num != desc_num);
+ return;
+ }
+
+ davinci_io_desc = desc;
+ davinci_io_desc_num = desc_num;
+
+ iotable_init(desc, desc_num);
+
+ /*
+ * Normally devicemaps_init() would flush caches and tlb after
+ * mdesc->map_io(), but we must also do it here because of the CPU
+ * revision check below.
This comments needs fixing as there's no revision check below -- it's
in different file now. :-)
+ */
+ local_flush_tlb_all();
+ flush_cache_all();
+}
WBR, Sergei
_______________________________________________
Davinci-linux-open-source mailing list
[email protected]
http://linux.davincidsp.com/mailman/listinfo/davinci-linux-open-source
