[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20191021133157.GA21112@linux-8ccs>
Date: Mon, 21 Oct 2019 15:31:57 +0200
From: Jessica Yu <jeyu@...nel.org>
To: Matthias Maennich <maennich@...gle.com>
Cc: linux-kernel@...r.kernel.org, kernel-team@...roid.com,
Masahiro Yamada <yamada.masahiro@...ionext.com>,
Martijn Coenen <maco@...roid.com>,
Lucas De Marchi <lucas.de.marchi@...il.com>,
Shaun Ruffell <sruffell@...ffell.net>,
Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
Will Deacon <will@...nel.org>, linux-kbuild@...r.kernel.org,
linux-modules@...r.kernel.org
Subject: Re: [PATCH v2 0/4] export/modpost: avoid renaming __ksymtab entries
for symbol namespaces
+++ Matthias Maennich [18/10/19 10:31 +0100]:
>The introduction of the symbol namespace patches changed the way symbols are
>named in the ksymtab entries. That caused userland tools to fail (such as
>kmod's depmod). As depmod is used as part of the kernel build it was worth
>having another look whether this name change can be avoided.
>
>The main purpose of this series is to restore the original ksymtab entry names.
>For that to happen and to remove some rough edges around that, the relevant
>parts in modpost got a small refactoring as to when and how namespaces are
>evaluated and set in the symbol struct.
>
>Eventually, the namespace values can be read from __kstrtabns_ entries and
>their corresponding __ksymtab_strings values. That removes the need to carry
>the namespace names within the (anyway unique) symbol name entries.
>
>The last patch of this series is adopted from Masahiro [1]. By allowing 'no
>namespace' to be represented as empty string, large chunks of
>include/linux/export.h could be consolidated. Technically, this last patch is
>not absolutely necessary to fix functionality. It addresses concerns about
>maintainability and readability. While I strongly suggest sending all of the
>patches for 5.4, the last one could possible deferred to the next merge window.
>
>This patch applies to the modules-linus [2] branch.
Hi!
I've applied the first three patches to modules-linus, inbound for -rc5.
Patch 4/4 was agreed to be for 5.5, and will be applied to
modules-next once I rebase that branch against -rc5.
Thanks!
Jessica
>Changes since v2:
> - restored correct authorship for [4/4]
> - add missing contributor tags
> - fixed typos and code style (spaces/tabs)
>
>[1] https://lore.kernel.org/lkml/20190927093603.9140-5-yamada.masahiro@socionext.com/
>[2] https://git.kernel.org/pub/scm/linux/kernel/git/jeyu/linux.git/log/?h=modules-linus
>
>Cc: Jessica Yu <jeyu@...nel.org>
>Cc: Masahiro Yamada <yamada.masahiro@...ionext.com>
>Cc: Martijn Coenen <maco@...roid.com>
>Cc: Lucas De Marchi <lucas.de.marchi@...il.com>
>Cc: Shaun Ruffell <sruffell@...ffell.net>
>Cc: Greg Kroah-Hartman <gregkh@...uxfoundation.org>
>Cc: Will Deacon <will@...nel.org>
>Cc: linux-kbuild@...r.kernel.org
>Cc: linux-modules@...r.kernel.org
>
>
>Masahiro Yamada (1):
> export: avoid code duplication in include/linux/export.h
>
>Matthias Maennich (3):
> modpost: delegate updating namespaces to separate function
> modpost: make updating the symbol namespace explicit
> symbol namespaces: revert to previous __ksymtab name scheme
>
> include/linux/export.h | 97 +++++++++++++-----------------------------
> kernel/module.c | 2 +-
> scripts/mod/modpost.c | 59 ++++++++++++++++---------
> scripts/mod/modpost.h | 1 +
> 4 files changed, 71 insertions(+), 88 deletions(-)
>
>Interdiff against v1:
>diff --git a/scripts/mod/modpost.c b/scripts/mod/modpost.c
>index 7cf0065ac95f..0bf7eab80d9f 100644
>--- a/scripts/mod/modpost.c
>+++ b/scripts/mod/modpost.c
>@@ -357,18 +357,21 @@ static const char *namespace_from_kstrtabns(struct elf_info *info,
>
> static void sym_update_namespace(const char *symname, const char *namespace)
> {
>- struct symbol *s = find_symbol(symname);
>- /* That symbol should have been created earlier and thus this is
>- * actually an assertion. */
>- if (!s) {
>- merror("Could not update namespace(%s) for symbol %s\n",
>- namespace, symname);
>- return;
>- }
>-
>- free(s->namespace);
>- s->namespace =
>- namespace && namespace[0] ? NOFAIL(strdup(namespace)) : NULL;
>+ struct symbol *s = find_symbol(symname);
>+
>+ /*
>+ * That symbol should have been created earlier and thus this is
>+ * actually an assertion.
>+ */
>+ if (!s) {
>+ merror("Could not update namespace(%s) for symbol %s\n",
>+ namespace, symname);
>+ return;
>+ }
>+
>+ free(s->namespace);
>+ s->namespace =
>+ namespace && namespace[0] ? NOFAIL(strdup(namespace)) : NULL;
> }
>
> /**
>--
>2.23.0.866.gb869b98d4c-goog
>
Powered by blists - more mailing lists