Masahiro Yamada <yamada.masahiro@xxxxxxxxxxxxx> writes:
> 2018-06-23 4:27 GMT+09:00 Dirk Gouders <dirk@xxxxxxxxxxx>:
>> xfwrite uses assert() to ensure it does not operate on empty strings.
>> Two users respect this, expr_print_file_helper() doesn't and causes
>> segfaults e.g. when the dependency for an empty default string is printed.
>>
>> Fix this by calling xfwrite with an empty string pattern ("") for
>> empty strings.
>>
>> INITRAMFS_SOURCE was one symbol that caused this problem, with this
>> fix applied the zconfdump() output for this symbol looks as follows:
>>
>> ------------------------------------------------------------------------
>> config INITRAMFS_SOURCE
>> string
>> symbol INITRAMFS_SOURCE
>> prompt "Initramfs source file(s)" if BLK_DEV_INITRD
>> default "" if BLK_DEV_INITRD
>> help
>> This can be either a single cpio archive with a .cpio suffix or a
>> space-separated list of directories and files for building the
>> initramfs image. A cpio archive should contain a filesystem archive
>> to be used as an initramfs image. Directories should contain a
>> filesystem layout to be included in the initramfs image. Files
>> should contain entries according to the format described by the
>> "usr/gen_init_cpio" program in the kernel tree.
>>
>> When multiple directories and files are specified then the
>> initramfs image will be the aggregate of all of them.
>>
>> See <file:Documentation/early-userspace/README> for more details.
>>
>> If you are not sure, leave it blank.
>> ------------------------------------------------------------------------
>>
>> Signed-off-by: Dirk Gouders <dirk@xxxxxxxxxxx>
>> ---
>
>
> Hmm, please let me think about this.
>
> I know this is an easy way to fix the seg-fault,
> but I think it is not a correct way.
>
>
>
> If the config entry were like this:
>
> config INITRAMFS_SOURCE
> string "Initramfs source file(s)"
> default "foo"
>
>
>
> It would be dumped like this:
>
> config INITRAMFS_SOURCE
> string
> symbol INITRAMFS_SOURCE
> prompt "Initramfs source file(s)"
> default foo
>
>
>
>
> I want it to be displayed like this:
>
> default "foo"
>
>
> If we can quote the string values,
> the problem will be eventually fixed.
>
> It may not be easy to fix it...
That's a good idea, it provides more consistency.
I will send an update and also replace the term "segfault" with "abort"
as this is the way assert() terminates the program.
Dirk
>
>> scripts/kconfig/expr.c | 5 ++++-
>> 1 file changed, 4 insertions(+), 1 deletion(-)
>>
>> diff --git a/scripts/kconfig/expr.c b/scripts/kconfig/expr.c
>> index e1a39e90841d..e064bf4c2881 100644
>> --- a/scripts/kconfig/expr.c
>> +++ b/scripts/kconfig/expr.c
>> @@ -1231,7 +1231,10 @@ void expr_print(struct expr *e,
>>
>> static void expr_print_file_helper(void *data, struct symbol *sym, const char *str)
>> {
>> - xfwrite(str, strlen(str), 1, data);
>> + if (*str != '\0')
>> + xfwrite(str, strlen(str), 1, data);
>> + else
>> + xfwrite("\"\"", 2, 1, data);
>> }
>>
>> void expr_fprint(struct expr *e, FILE *out)
>> --
>> 2.13.6
>>
>> --
>> To unsubscribe from this list: send the line "unsubscribe linux-kbuild" in
>> the body of a message to majordomo@xxxxxxxxxxxxxxx
>> More majordomo info at http://vger.kernel.org/majordomo-info.html
--
To unsubscribe from this list: send the line "unsubscribe linux-kbuild" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
