[U-Boot] [PATCH 1/7] fdt_support: Add multi-serial support for stdout fixup

Currently fdt_fixup_stdout() is using hard-coded CONFIG_CONS_INDEX constant. With multi-serial support, the CONS_INDEX may no longer represent actual console, so we should try to extract port number from the current stdio device name instead of always hard-coding the constant value.
Signed-off-by: Anton Vorontsov avorontsov@ru.mvista.com --- common/fdt_support.c | 22 +++++++++++++++++++++- 1 files changed, 21 insertions(+), 1 deletions(-)
diff --git a/common/fdt_support.c b/common/fdt_support.c index 89164a1..e01303a 100644 --- a/common/fdt_support.c +++ b/common/fdt_support.c @@ -22,6 +22,7 @@ */
#include <common.h> +#include <stdio_dev.h> #include <linux/ctype.h> #include <linux/types.h> #include <asm/global_data.h> @@ -90,6 +91,23 @@ int fdt_find_and_setprop(void *fdt, const char *node, const char *prop, }
#ifdef CONFIG_OF_STDOUT_VIA_ALIAS + +#ifdef CONFIG_SERIAL_MULTI +static void fdt_fill_multisername(char *sername, size_t maxlen) +{ + const char *outname = stdio_devices[stdout]->name; + + if (strcmp(outname, "serial") > 0) + strncpy(sername, outname, maxlen); + + /* eserial? */ + if (strcmp(outname + 1, "serial") > 0) + strncpy(sername, outname + 1, maxlen); +} +#else +static inline void fdt_fill_multisername(char *sername, size_t maxlen) {} +#endif /* CONFIG_SERIAL_MULTI */ + static int fdt_fixup_stdout(void *fdt, int chosenoff) { int err = 0; @@ -98,7 +116,9 @@ static int fdt_fixup_stdout(void *fdt, int chosenoff) char sername[9] = { 0 }; const char *path;
- sprintf(sername, "serial%d", CONFIG_CONS_INDEX - 1); + fdt_fill_multisername(sername, sizeof(sername) - 1); + if (!sername[0]) + sprintf(sername, "serial%d", CONFIG_CONS_INDEX - 1);
err = node = fdt_path_offset(fdt, "/aliases"); if (node >= 0) {

On Aug 19, 2009, at 1:37 PM, Anton Vorontsov wrote:
Currently fdt_fixup_stdout() is using hard-coded CONFIG_CONS_INDEX constant. With multi-serial support, the CONS_INDEX may no longer represent actual console, so we should try to extract port number from the current stdio device name instead of always hard-coding the constant value.
Signed-off-by: Anton Vorontsov avorontsov@ru.mvista.com
common/fdt_support.c | 22 +++++++++++++++++++++- 1 files changed, 21 insertions(+), 1 deletions(-)
Should get some Ack from Jerry.
- k

Hi Kumar, Anton,
Kumar Gala wrote:
On Aug 19, 2009, at 1:37 PM, Anton Vorontsov wrote:
Currently fdt_fixup_stdout() is using hard-coded CONFIG_CONS_INDEX constant. With multi-serial support, the CONS_INDEX may no longer represent actual console, so we should try to extract port number from the current stdio device name instead of always hard-coding the constant value.
Signed-off-by: Anton Vorontsov avorontsov@ru.mvista.com
common/fdt_support.c | 22 +++++++++++++++++++++- 1 files changed, 21 insertions(+), 1 deletions(-)
Should get some Ack from Jerry.
- k
Sorry, I was asleep at the wheel.
I'm OK with the patch as is, but it could be tightened up slightly by having the new function fill in the serial name in the non-multi situation. I renamed the function as well to be fdt_fill_sername() to remove the "multi" connotation.
Below is a *HAND MODIFIED* (i.e. probably broken) edit of Anton's patch reflecting my thoughts.
I'll leave it up to Anton and Kumar... I'm OK with Anton's original patch and you have my Acked-by: Gerald Van Baren vanbaren@cideas.com or you can verify and adopt my proposed change (more work :-() and add a Signed-off-by: Gerald Van Baren vanbaren@cideas.com to Anton's s-o-b.
Thanks, gvb ------------------------------------------------------------------------
Currently fdt_fixup_stdout() is using hard-coded CONFIG_CONS_INDEX constant. With multi-serial support, the CONS_INDEX may no longer represent actual console, so we should try to extract port number from the current stdio device name instead of always hard-coding the constant value.
Signed-off-by: Anton Vorontsov avorontsov@ru.mvista.com Signed-off-by: Gerald Van Baren vanbaren@cideas.com --- common/fdt_support.c | 22 +++++++++++++++++++++- 1 files changed, 21 insertions(+), 1 deletions(-)
diff --git a/common/fdt_support.c b/common/fdt_support.c index 89164a1..e01303a 100644 --- a/common/fdt_support.c +++ b/common/fdt_support.c @@ -22,6 +22,7 @@ */
#include <common.h> +#include <stdio_dev.h> #include <linux/ctype.h> #include <linux/types.h> #include <asm/global_data.h> @@ -90,6 +91,26 @@ int fdt_find_and_setprop(void *fdt, const char *node, const char *prop, }
#ifdef CONFIG_OF_STDOUT_VIA_ALIAS + +#ifdef CONFIG_SERIAL_MULTI +static void fdt_fill_sername(char *sername, size_t maxlen) +{ + const char *outname = stdio_devices[stdout]->name; + + if (strcmp(outname, "serial") > 0) + strncpy(sername, outname, maxlen); + + /* eserial? */ + if (strcmp(outname + 1, "serial") > 0) + strncpy(sername, outname + 1, maxlen); +} +#else +static inline void fdt_fill_sername(char *sername, size_t maxlen) +{ + sprintf(sername, "serial%d", CONFIG_CONS_INDEX - 1); +} +#endif /* CONFIG_SERIAL_MULTI */ + static int fdt_fixup_stdout(void *fdt, int chosenoff) { int err = 0; @@ -98,7 +116,7 @@ static int fdt_fixup_stdout(void *fdt, int chosenoff) char sername[9] = { 0 }; const char *path;
- sprintf(sername, "serial%d", CONFIG_CONS_INDEX - 1); + fdt_fill_sername(sername, sizeof(sername) - 1);
err = node = fdt_path_offset(fdt, "/aliases"); if (node >= 0) {

Thanks for review Jerry.
On Wed, Sep 16, 2009 at 08:59:14AM -0400, Jerry Van Baren wrote: [...]
Below is a *HAND MODIFIED* (i.e. probably broken) edit of Anton's patch reflecting my thoughts.
I'll leave it up to Anton and Kumar... I'm OK with Anton's original patch and you have my Acked-by: Gerald Van Baren vanbaren@cideas.com or you can verify and adopt my proposed change (more work :-() and add a Signed-off-by: Gerald Van Baren vanbaren@cideas.com to Anton's s-o-b.
Thanks, gvb
[...]
#ifdef CONFIG_OF_STDOUT_VIA_ALIAS
+#ifdef CONFIG_SERIAL_MULTI +static void fdt_fill_sername(char *sername, size_t maxlen) +{
- const char *outname = stdio_devices[stdout]->name;
- if (strcmp(outname, "serial") > 0)
strncpy(sername, outname, maxlen);
(1)
- /* eserial? */
- if (strcmp(outname + 1, "serial") > 0)
strncpy(sername, outname + 1, maxlen);
+} +#else +static inline void fdt_fill_sername(char *sername, size_t maxlen) +{
- sprintf(sername, "serial%d", CONFIG_CONS_INDEX - 1);
This case is also needed with multiserial, since 'serial' is what you get by default, and (1) doesn't account 'serial' name.
So, we can either leave the patch as, or duplicate some code for multiserial case, or do this ugly thing (I quite dislike #ifdef like this, i.e. inside the code flow):
static void fdt_fill_sername(char *sername, size_t maxlen) { #ifdef CONFIG_SERIAL_MULTI const char *outname = stdio_devices[stdout]->name;
if (strcmp(outname, "serial") > 0) strncpy(sername, outname, maxlen);
/* eserial? */ if (strcmp(outname + 1, "serial") > 0) strncpy(sername, outname + 1, maxlen); #endif if (!sername[0]) sprintf(sername, "serial%d", CONFIG_CONS_INDEX - 1); }
Though, I'll do whatever you prefer.
Thanks,

Anton Vorontsov wrote:
Thanks for review Jerry.
On Wed, Sep 16, 2009 at 08:59:14AM -0400, Jerry Van Baren wrote: [...]
Below is a *HAND MODIFIED* (i.e. probably broken) edit of Anton's patch reflecting my thoughts.
I'll leave it up to Anton and Kumar... I'm OK with Anton's original patch and you have my Acked-by: Gerald Van Baren vanbaren@cideas.com or you can verify and adopt my proposed change (more work :-() and add a Signed-off-by: Gerald Van Baren vanbaren@cideas.com to Anton's s-o-b.
Thanks, gvb
[...]
#ifdef CONFIG_OF_STDOUT_VIA_ALIAS
+#ifdef CONFIG_SERIAL_MULTI +static void fdt_fill_sername(char *sername, size_t maxlen) +{
- const char *outname = stdio_devices[stdout]->name;
- if (strcmp(outname, "serial") > 0)
strncpy(sername, outname, maxlen);
(1)
- /* eserial? */
- if (strcmp(outname + 1, "serial") > 0)
strncpy(sername, outname + 1, maxlen);
+} +#else +static inline void fdt_fill_sername(char *sername, size_t maxlen) +{
- sprintf(sername, "serial%d", CONFIG_CONS_INDEX - 1);
This case is also needed with multiserial, since 'serial' is what you get by default, and (1) doesn't account 'serial' name.
Ahh, whups, missed that.
So, we can either leave the patch as, or duplicate some code for multiserial case, or do this ugly thing (I quite dislike #ifdef like this, i.e. inside the code flow):
[snip]
Though, I'll do whatever you prefer.
Lets use the original patch.
Acked-by: Gerald Van Baren vanbaren@cideas.com
Thanks, gvb
participants (3)
-
Anton Vorontsov
-
Jerry Van Baren
-
Kumar Gala