summaryrefslogtreecommitdiffstats
path: root/Documentation/security
diff options
context:
space:
mode:
authorJulius Werner <jwerner@chromium.org>2021-02-24 16:41:55 -0800
committerJulius Werner <jwerner@chromium.org>2021-02-26 00:49:35 +0000
commit7f0d3de251f60dd71f48f87457c3057ec7a71148 (patch)
tree126e1b0b2592ba79eed3b4f8a5987523b296db7c /Documentation/security
parent6e3f38392720bbedd0d493ec61c9d211a36670a8 (diff)
downloadcoreboot-7f0d3de251f60dd71f48f87457c3057ec7a71148.tar.gz
coreboot-7f0d3de251f60dd71f48f87457c3057ec7a71148.tar.bz2
coreboot-7f0d3de251f60dd71f48f87457c3057ec7a71148.zip
rk3399: sdram: Remove obs_err variable to avoid uninitialized use
CB:50863 refactored the data_training() function to split out read gate training into a separate function, but in the course of this forgot to correctly initialize the local obs_err varible in the new function to 0. This means that it will be used uninitialized, and when it happens to be non-zero it makes the training process fail. Due to the convoluted control flow in the function, it seems that GCC's static analyzer couldn't pick up on this uninitialized use. The whole variable is unnecessary anyway, all it's used for is to force the function to return two lines below without doing anything with side-effects in between. This patch removes the variable and simplifies the code in all three training functions to avoid this uninitialized use issue and make everything a bit more readable. (Also restore the original pre-clang-format continuation line intendations for more readability.) Signed-off-by: Julius Werner <jwerner@chromium.org> Change-Id: Ia475d64c06f2ec1bf9295742d173ce66717b821c Reviewed-on: https://review.coreboot.org/c/coreboot/+/51079 Reviewed-by: Moritz Fischer <moritzf@google.com> Reviewed-by: Angel Pons <th3fanbus@gmail.com> Tested-by: build bot (Jenkins) <no-reply@coreboot.org>
Diffstat (limited to 'Documentation/security')
0 files changed, 0 insertions, 0 deletions