Skip to contentSkip to conversation

2.6 zaurus: pcmcia now works

DISCUSSION Pavel Machek 3 participants 3 messages Sep 20, 2005 Multifunction devices (MFD)

lore
Share & tools
lore
#0 lore raw

Hi!

With great help from Richard Purdie, pcmcia now works on sl-5500. Thanks! I also updated to 2.6.14-rc1, and it still seems to work :-).

FYI, my hacks now look like (attached). I'll push changes to linux-z in few hours.

Pavel

Signature · 1 line
if you have sharp zaurus hardware you don't need... you know my address
#1 lore raw
Quoted 2 lines · Pavel Machek

FYI, my hacks now look like (attached). I'll push changes to linux-z in few hours.

Some questions:

@@ -57,8 +57,6 @@ void dump_backtrace_entry(unsigned long

 #ifdef CONFIG_KALLSYMS
        printk("[<%08lx>] ", where);
        print_symbol("(%s) ", where);

- printk("from [<%08lx>] ", from);
- print_symbol("(%s)\n", from);

The "from" address provides good hints about the exact path we got to the called function. You don't really want to get rid of that because it makes following backtraces harder. I'm not sure why you've made the other changes in that file either.

+/* those must never be empty
+ unfortunately they cause problems with older binutils
ASSERT((__proc_info_end - __proc_info_begin), "missing CPU support") ASSERT((__arch_info_end - __arch_info_begin), "no machine record defined")
+*/

Get a better binutils. 8)

drivers/mfd/mcp-core.c +1 -0
@@ -19,6 +19,7 @@
 #include <asm/dma.h>
 #include <asm/system.h>
 
+#include <asm/arch/mcp.h>
 #include "mcp.h"
 
 #define to_mcp(d)              container_of(d, struct mcp, attached_device)

This looks bogus - why is this needed?

@@ -186,7 +192,12 @@ static int mcp_sa11x0_probe(struct devic

         */
        Ser4MCSR = -1;
        Ser4MCCR1 = data->mccr1;

- Ser4MCCR0 = data->mccr0 | 0x7f7f;
+#if 1
+ if (machine_is_collie())
+ Ser4MCCR0 = MCCR0_ADM | MCCR0_ExtClk;
+ else
+#endif
+ Ser4MCCR0 = data->mccr0 | 0x7f7f;

Ditto.

Signature · 3 lines
Russell King
 Linux kernel    2.6 ARM Linux   - http://www.arm.linux.org.uk/
 maintainer of:  2.6 Serial core
#2 lore raw

Hi!

Quoted 12 lines
Quoted 2 lines

FYI, my hacks now look like (attached). I'll push changes to linux-z in few hours.

Some questions:

(quoted diff) +0 -2
@@ -57,8 +57,6 @@ void dump_backtrace_entry(unsigned long
 #ifdef CONFIG_KALLSYMS
        printk("[<%08lx>] ", where);
        print_symbol("(%s) ", where);
-       printk("from [<%08lx>] ", from);
-       print_symbol("(%s)\n", from);
 
Quoted 4 lines

The "from" address provides good hints about the exact path we got to the called function. You don't really want to get rid of that because it makes following backtraces harder. I'm not sure why you've made the other changes in that file either.

Well, zaurus console is something like 40x20 characters. Normal backtrace does not fit there, and it has most important info at the top.

Quoted 7 lines
(quoted diff) +3 -0
+/* those must never be empty
+   unfortunately they cause problems with older binutils
 ASSERT((__proc_info_end - __proc_info_begin), "missing CPU support")
 ASSERT((__arch_info_end - __arch_info_begin), "no machine record defined")
+*/

Get a better binutils. 8)

Yep :-). Those can wait.

Quoted 12 lines
drivers/mfd/mcp-core.c +1 -0
@@ -19,6 +19,7 @@
 #include <asm/dma.h>
 #include <asm/system.h>
 
+#include <asm/arch/mcp.h>
 #include "mcp.h"
 
 #define to_mcp(d)              container_of(d, struct mcp, attached_device)
Quoted 1 line

This looks bogus - why is this needed?

Can go, fixed. (Remainings of previous patches).

Quoted 12 lines
drivers/mfd/mcp-core.c +6 -1
@@ -186,7 +192,12 @@ static int mcp_sa11x0_probe(struct devic
         */
        Ser4MCSR = -1;
        Ser4MCCR1 = data->mccr1;
-       Ser4MCCR0 = data->mccr0 | 0x7f7f;
+#if 1
+       if (machine_is_collie())
+               Ser4MCCR0 = MCCR0_ADM | MCCR0_ExtClk;
+       else
+#endif
+               Ser4MCCR0 = data->mccr0 | 0x7f7f;
 
Quoted 1 line

Ditto.

I tried to kill that one before, and it broke boot. I have little idea how to debug that one...
Pavel

Signature · 1 line
Boycott Kodak -- for their patent abuse against Java.