[PATCH] scripts: decode_stacktrace: preserve input while reading lines

[PATCH] scripts: decode_stacktrace: preserve input while reading lines

[Hamdy Abou El Anein]

From: Hamdy Abou El Anein <hamdy.aea@protonmail.com>

Plain read strips leading and trailing whitespace and interprets
backslashes before the decoder sees a line. It also drops a final line
without a newline. This loses log text and defeats the alignment
preservation in handle_line().

Read with an empty IFS and -r, and process a nonempty final fragment.

Tested passthrough text with indentation, backslashes and trailing
spaces, plus a real debug-enabled object and stack entries terminated
by LF, CRLF or EOF. All three stack entries decode with their original
indentation.

AI assistance: ChatGPT/Codex identified the issue and prepared the
change, commit message and executable regression tests.

Fixes: dbd1abb20971 ("decode_stacktrace: make stack dump output useful again")
Signed-off-by: Hamdy Abou El Anein <hamdy.aea@protonmail.com>
---
 scripts/decode_stacktrace.sh | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/scripts/decode_stacktrace.sh b/scripts/decode_stacktrace.sh
index 39d60d477bf..8b06810d463 100755
--- a/scripts/decode_stacktrace.sh
+++ b/scripts/decode_stacktrace.sh
@@ -346,7 +346,7 @@ handle_line() {
 	echo "${spaces[$last]}${symbol}${module:+ ${module}}${info_str:+ ${info_str}}"
 }
 
-while read line; do
+while IFS= read -r line || [[ -n $line ]]; do
 	# Strip unexpected carriage return at end of line
 	line=${line%$'\r'}
 

Re: [PATCH] scripts: decode_stacktrace: preserve input while reading lines

[Sasha Levin]

On Tue, Nov 03, 2026 at 05:42:03PM +0100, Hamdy Abou El Anein wrote:
>Plain read strips leading and trailing whitespace and interprets
>backslashes before the decoder sees a line. It also drops a final line
>without a newline. This loses log text and defeats the alignment
>preservation in handle_line().
>
>Read with an empty IFS and -r, and process a nonempty final fragment.

One optional follow-up: debuginfod_get_vmlinux() takes the build ID with ${1##*
}, so trailing whitespace after the build ID now gives an empty ID and the
lookup fails. The kernel never prints that, since the build ID always ends the
line, so it only matters for logs edited after the fact.

Acked-by: Sasha Levin <sashal@kernel.org>

-- 
Thanks,
Sasha