| Message ID | 20230808-arch-um-v2-1-5fbe76eaf853@google.com (mailing list archive) |
|---|---|
| State | Changes Requested |
| Headers | show |
| Series | [v2] um: refactor deprecated strncpy to strtomem | expand |
On Tue, Aug 08, 2023 at 09:17:35PM +0000, Justin Stitt wrote: > Favor a more robust and less ambiguous interface over `strncpy`. > > Use `strtomem` here since `console_buf` is not expected to be > NUL-terminated. > > Also mark char buffer as `__nonstring` as per Kees' suggestion here [1] > > Link: https://github.com/KSPP/linux/issues/90 [1] > Link: www.kernel.org/doc/html/latest/process/deprecated.html#strncpy-on-nul-terminated-strings > Cc: linux-hardening@vger.kernel.org > Signed-off-by: Justin Stitt <justinstitt@google.com> > --- > Changes in v2: > - Keep usage of ARRAY_SIZE (thanks Bill) > - Remove unnecessary import (thanks Kees and Bill) > - Link to v1: https://lore.kernel.org/r/20230807-arch-um-v1-1-86dbbfb59709@google.com > --- > Notes: > I only build tested this patch. > > I am also making an assumption that `console_buf` need not be > NUL-terminated. Am I right in believing this? > --- > arch/um/drivers/mconsole_kern.c | 4 ++-- > 1 file changed, 2 insertions(+), 2 deletions(-) > > diff --git a/arch/um/drivers/mconsole_kern.c b/arch/um/drivers/mconsole_kern.c > index 5026e7b9adfe..d5fa9c8dcd18 100644 > --- a/arch/um/drivers/mconsole_kern.c > +++ b/arch/um/drivers/mconsole_kern.c > @@ -554,7 +554,7 @@ struct mconsole_output { > > static DEFINE_SPINLOCK(client_lock); > static LIST_HEAD(clients); > -static char console_buf[MCONSOLE_MAX_DATA]; > +static char console_buf[MCONSOLE_MAX_DATA] __nonstring; > > static void console_write(struct console *console, const char *string, > unsigned int len) > @@ -567,7 +567,7 @@ static void console_write(struct console *console, const char *string, > > while (len > 0) { > n = min((size_t) len, ARRAY_SIZE(console_buf)); > - strncpy(console_buf, string, n); > + strtomem(console_buf, string); See my reply to v1: I think this should just be memcpy(console_buf, string, n) -Kees > string += n; > len -= n; > > > --- > base-commit: c1a515d3c0270628df8ae5f5118ba859b85464a2 > change-id: 20230807-arch-um-3ef24413427e > > Best regards, > -- > Justin Stitt <justinstitt@google.com> >
diff --git a/arch/um/drivers/mconsole_kern.c b/arch/um/drivers/mconsole_kern.c index 5026e7b9adfe..d5fa9c8dcd18 100644 --- a/arch/um/drivers/mconsole_kern.c +++ b/arch/um/drivers/mconsole_kern.c @@ -554,7 +554,7 @@ struct mconsole_output { static DEFINE_SPINLOCK(client_lock); static LIST_HEAD(clients); -static char console_buf[MCONSOLE_MAX_DATA]; +static char console_buf[MCONSOLE_MAX_DATA] __nonstring; static void console_write(struct console *console, const char *string, unsigned int len) @@ -567,7 +567,7 @@ static void console_write(struct console *console, const char *string, while (len > 0) { n = min((size_t) len, ARRAY_SIZE(console_buf)); - strncpy(console_buf, string, n); + strtomem(console_buf, string); string += n; len -= n;
Favor a more robust and less ambiguous interface over `strncpy`. Use `strtomem` here since `console_buf` is not expected to be NUL-terminated. Also mark char buffer as `__nonstring` as per Kees' suggestion here [1] Link: https://github.com/KSPP/linux/issues/90 [1] Link: www.kernel.org/doc/html/latest/process/deprecated.html#strncpy-on-nul-terminated-strings Cc: linux-hardening@vger.kernel.org Signed-off-by: Justin Stitt <justinstitt@google.com> --- Changes in v2: - Keep usage of ARRAY_SIZE (thanks Bill) - Remove unnecessary import (thanks Kees and Bill) - Link to v1: https://lore.kernel.org/r/20230807-arch-um-v1-1-86dbbfb59709@google.com --- Notes: I only build tested this patch. I am also making an assumption that `console_buf` need not be NUL-terminated. Am I right in believing this? --- arch/um/drivers/mconsole_kern.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) --- base-commit: c1a515d3c0270628df8ae5f5118ba859b85464a2 change-id: 20230807-arch-um-3ef24413427e Best regards, -- Justin Stitt <justinstitt@google.com>