[U-Boot] [PATCH 1/2] libfdt: overlay: Fix missing symbols condition

From: Stefan Agner stefan.agner@toradex.com
When there is no symbols section in the device tree, overlay_fixup_phandles should return FDT_ERR_NOTFOUND instead of FDT_ERR_BADOFFSET.
Signed-off-by: Stefan Agner stefan.agner@toradex.com ---
lib/libfdt/fdt_overlay.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/lib/libfdt/fdt_overlay.c b/lib/libfdt/fdt_overlay.c index bb41404..4a9ba40 100644 --- a/lib/libfdt/fdt_overlay.c +++ b/lib/libfdt/fdt_overlay.c @@ -495,9 +495,9 @@ static int overlay_fixup_phandles(void *fdt, void *fdto) if ((fixups_off < 0 && (fixups_off != -FDT_ERR_NOTFOUND))) return fixups_off;
- /* And base DTs without symbols */ + /* But if we need to fixup phandles, symbols are required */ symbols_off = fdt_path_offset(fdt, "/__symbols__"); - if ((symbols_off < 0 && (symbols_off != -FDT_ERR_NOTFOUND))) + if (symbols_off < 0) return symbols_off;
fdt_for_each_property_offset(property, fdto, fixups_off) {

From: Stefan Agner stefan.agner@toradex.com
There are lots of reason why a FDT application might fail, the error code might give an indication. Let the error code translate in a error string so users can try to understand what went wrong.
Signed-off-by: Stefan Agner stefan.agner@toradex.com --- Currently I get this when I apply a device tree overlay to a device tree without symbols:
Colibri iMX7 # fdt apply ${loadaddr} Colibri iMX7 # fdt print libfdt fdt_path_offset() returned FDT_ERR_BADMAGIC Colibri iMX7 #
With this change, the situation is a bit more indicative:
Colibri iMX7 # fdt apply ${loadaddr} fdt_overlay_apply(): FDT_ERR_NOTFOUND Colibri iMX7 # fdt print libfdt fdt_path_offset() returned FDT_ERR_BADMAGIC Colibri iMX7 #
cmd/fdt.c | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-)
diff --git a/cmd/fdt.c b/cmd/fdt.c index 8bd345a..6883e75 100644 --- a/cmd/fdt.c +++ b/cmd/fdt.c @@ -642,6 +642,7 @@ static int do_fdt(cmd_tbl_t *cmdtp, int flag, int argc, char * const argv[]) else if (strncmp(argv[1], "ap", 2) == 0) { unsigned long addr; struct fdt_header *blob; + int ret;
if (argc != 3) return CMD_RET_USAGE; @@ -654,8 +655,11 @@ static int do_fdt(cmd_tbl_t *cmdtp, int flag, int argc, char * const argv[]) if (!fdt_valid(&blob)) return CMD_RET_FAILURE;
- if (fdt_overlay_apply(working_fdt, blob)) + ret = fdt_overlay_apply(working_fdt, blob); + if (ret) { + printf("fdt_overlay_apply(): %s\n", fdt_strerror(ret)); return CMD_RET_FAILURE; + } } #endif /* resize the fdt */

On Thu, Dec 15, 2016 at 03:03:27PM -0800, Stefan Agner wrote:
From: Stefan Agner stefan.agner@toradex.com
There are lots of reason why a FDT application might fail, the error code might give an indication. Let the error code translate in a error string so users can try to understand what went wrong.
Signed-off-by: Stefan Agner stefan.agner@toradex.com
Acked-by: Maxime Ripard maxime.ripard@free-electrons.com
Thanks! Maxime

On 16 December 2016 at 02:33, Maxime Ripard maxime.ripard@free-electrons.com wrote:
On Thu, Dec 15, 2016 at 03:03:27PM -0800, Stefan Agner wrote:
From: Stefan Agner stefan.agner@toradex.com
There are lots of reason why a FDT application might fail, the error code might give an indication. Let the error code translate in a error string so users can try to understand what went wrong.
Signed-off-by: Stefan Agner stefan.agner@toradex.com
Acked-by: Maxime Ripard maxime.ripard@free-electrons.com
Acked-by: Simon Glass sjg@chromium.org

On 17 December 2016 at 15:48, Simon Glass sjg@chromium.org wrote:
On 16 December 2016 at 02:33, Maxime Ripard maxime.ripard@free-electrons.com wrote:
On Thu, Dec 15, 2016 at 03:03:27PM -0800, Stefan Agner wrote:
From: Stefan Agner stefan.agner@toradex.com
There are lots of reason why a FDT application might fail, the error code might give an indication. Let the error code translate in a error string so users can try to understand what went wrong.
Signed-off-by: Stefan Agner stefan.agner@toradex.com
Acked-by: Maxime Ripard maxime.ripard@free-electrons.com
Acked-by: Simon Glass sjg@chromium.org
Applied to u-boot-fdt, thanks!

On Thu, Dec 15, 2016 at 03:03:26PM -0800, Stefan Agner wrote:
From: Stefan Agner stefan.agner@toradex.com
When there is no symbols section in the device tree, overlay_fixup_phandles should return FDT_ERR_NOTFOUND instead of FDT_ERR_BADOFFSET.
Signed-off-by: Stefan Agner stefan.agner@toradex.com
lib/libfdt/fdt_overlay.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/lib/libfdt/fdt_overlay.c b/lib/libfdt/fdt_overlay.c index bb41404..4a9ba40 100644 --- a/lib/libfdt/fdt_overlay.c +++ b/lib/libfdt/fdt_overlay.c @@ -495,9 +495,9 @@ static int overlay_fixup_phandles(void *fdt, void *fdto) if ((fixups_off < 0 && (fixups_off != -FDT_ERR_NOTFOUND))) return fixups_off;
- /* And base DTs without symbols */
- /* But if we need to fixup phandles, symbols are required */ symbols_off = fdt_path_offset(fdt, "/__symbols__");
- if ((symbols_off < 0 && (symbols_off != -FDT_ERR_NOTFOUND)))
- if (symbols_off < 0) return symbols_off;
A base device tree doesn't need to have a symbols node. In the (uncommon, true, but still real) case where you wouldn't have any label in the base DT, __symbols__ wouldn't be there.
Now, that code would return an error only if there's such a reference expressed in __fixups__, but no __symbols__ node that also has that reference. This is definitely an error but not really a NOTFOUND. You had an offset already, but this offset was bad. Hence the error.
Do you have a base DT and an overlay showing up the error you're trying to fix? Either way, that should be discussed with upstream's libfdt ml and maintainer, not (only) on U-Boot.
Maxime

On 2016-12-16 01:32, Maxime Ripard wrote:
On Thu, Dec 15, 2016 at 03:03:26PM -0800, Stefan Agner wrote:
From: Stefan Agner stefan.agner@toradex.com
When there is no symbols section in the device tree, overlay_fixup_phandles should return FDT_ERR_NOTFOUND instead of FDT_ERR_BADOFFSET.
Signed-off-by: Stefan Agner stefan.agner@toradex.com
lib/libfdt/fdt_overlay.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/lib/libfdt/fdt_overlay.c b/lib/libfdt/fdt_overlay.c index bb41404..4a9ba40 100644 --- a/lib/libfdt/fdt_overlay.c +++ b/lib/libfdt/fdt_overlay.c @@ -495,9 +495,9 @@ static int overlay_fixup_phandles(void *fdt, void *fdto) if ((fixups_off < 0 && (fixups_off != -FDT_ERR_NOTFOUND))) return fixups_off;
- /* And base DTs without symbols */
- /* But if we need to fixup phandles, symbols are required */ symbols_off = fdt_path_offset(fdt, "/__symbols__");
- if ((symbols_off < 0 && (symbols_off != -FDT_ERR_NOTFOUND)))
- if (symbols_off < 0) return symbols_off;
A base device tree doesn't need to have a symbols node. In the (uncommon, true, but still real) case where you wouldn't have any label in the base DT, __symbols__ wouldn't be there.
Yeah I know, but in case there are fixup offsets found (just above), symbols get essentially mandatory.
Now, that code would return an error only if there's such a reference expressed in __fixups__, but no __symbols__ node that also has that reference. This is definitely an error but not really a NOTFOUND. You had an offset already, but this offset was bad. Hence the error.
It probably would a new error code?
Do you have a base DT and an overlay showing up the error you're trying to fix? Either way, that should be discussed with upstream's libfdt ml and maintainer, not (only) on U-Boot.
Yeah my problem was that my base DT had no symbols and used labels. Wasn't aware of the whole symbols problematic...
I actually just realized that this got fixed upstream: https://git.kernel.org/cgit/utils/dtc/dtc.git/commit/libfdt/fdt_overlay.c?id...
I guess we could backport that fix?
-- Stefan

On Sat, Dec 17, 2016 at 04:56:49PM -0800, Stefan Agner wrote:
Do you have a base DT and an overlay showing up the error you're trying to fix? Either way, that should be discussed with upstream's libfdt ml and maintainer, not (only) on U-Boot.
Yeah my problem was that my base DT had no symbols and used labels. Wasn't aware of the whole symbols problematic...
I actually just realized that this got fixed upstream: https://git.kernel.org/cgit/utils/dtc/dtc.git/commit/libfdt/fdt_overlay.c?id...
I guess we could backport that fix?
Yep.
Thanks! Maxime
participants (3)
-
Maxime Ripard
-
Simon Glass
-
Stefan Agner