Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::3:3 as permitted sender) client-ip=2620:137:e000::3:3;
Date:   Thu, 16 Nov 2023 14:28:57 +0200
From:   Boris Kolpackov <boris@codesynthesis.com>
To:     Masahiro Yamada <masahiroy@kernel.org>
Cc:     linux-kbuild@vger.kernel.org, linux-kernel@vger.kernel.org
Subject: Re: [PATCH] kconfig: fix memory leak from range properties
Message-ID: <boris.20231116141906@codesynthesis.com>
References: <20231115041653.810045-1-masahiroy@kernel.org>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20231115041653.810045-1-masahiroy@kernel.org>
Organization: Code Synthesis
Precedence: bulk

Masahiro Yamada <masahiroy@kernel.org> writes:

> Currently, sym_validate_range() duplicates the range string using
> xstrdup(), which is overwritten by a subsequent sym_calc_value() call.
> It results in a memory leak.
> 
> [After]
> 
>   LEAK SUMMARY:
>      definitely lost: 0 bytes in 0 blocks
>      indirectly lost: 0 bytes in 0 blocks
>        possibly lost: 0 bytes in 0 blocks
>      still reachable: 17,462 bytes in 20 blocks
>           suppressed: 0 bytes in 0 blocks

FYI, there are quite a few other memory leaks in Kconfig (as evident from
the still reachable value in the above report). I believe I've fixed most
of them in this commit:

https://github.com/build2-packaging/kconfig/commit/cd9910e3636515b2980ce1d37d1984ccfd6b4cb9

In particular, I could load the Linux kernel configuration repeatedly
in a loop without causing any memory leaks or crashes (which were common
due to the state not being reset properly).

I believe the above commit also includes a fix for the sym_validate_range()
leak in question, though the way it's fixed is different. The potential
problem with the proposed fix is that it may be impossible to decide who
should free the shared value.