Dear Patrick,
In message 1512575263-23010-1-git-send-email-patrick.delaunay@st.com you wrote:
In function get_ram_size() and for 2 last cases the content of the base address (*base) is not restored even it is correctly saved in stack (in save[i]).
This patch solved this issue. The content of the base address is saved in new variable in stack (save_base) to avoid the need of other information (value of i) and restored in all the cases.
What exactly is the problem you are trying to fix? How exactly does it manifest for you?
On which boards/architectures did you observe this problem, and on which did you actually test your patch?
How exactly is your memory mapped and tested on the boards where your patch fixes a problem?
The thing is, that this "fix" comes up again and again wevery coplu of months / years, and IIRC so far all these patches broke some system, while the code as is has been working fine of many systems.
See for example commit b8496cce and revert in 3ab270d5 in 2012, or commit 8e7cba04 and revert in cc8d698f in 2016.
See also the threads starting at
Subject: [U-Boot-Users] memsize.c patch From: "Sangmoon Kim" dogoil@etinsys.com Date: Fri, 2 Apr 2004 13:08:50 +0900
and
Subject: [PATCH v2] memsize: Fix for bug in memory sizing code From: Gerd Hoffmann kraxel@redhat.com Date: Tue, 21 Oct 2014 18:49:13 +0200
Thanks.
Best regards,
Wolfgang Denk