This serie fixes an off by one related to the usage of strncat() in
octeon_download_firmware().
The first patch is a minimal fix.
The 2nd one is an attempt to remove strncat() which is used in a wrong way
most of the time.
It removes the need of an intermediate buffer but may need further discussions.
(i.e. is it a good idea to update h->bootcmd directly?)
Both patches are compile tested only.
Christophe JAILLET (2):
liquidio: Fix an off by one in octeon_download_firmware()
liquidio: Simplify octeon_download_firmware()
.../net/ethernet/cavium/liquidio/octeon_console.c | 13 +++++--------
1 file changed, 5 insertions(+), 8 deletions(-)
--
2.34.1
In order to remove the usage of strncat(), write directly at the rigth
place in the 'h->bootcmd' array and check if the output is truncated.
Signed-off-by: Christophe JAILLET <[email protected]>
---
The goal is to potentially remove the strncat() function from the kernel.
Their are only few users and most of them use it wrongly.
This patch is compile tested only.
---
.../net/ethernet/cavium/liquidio/octeon_console.c | 13 +++++--------
1 file changed, 5 insertions(+), 8 deletions(-)
diff --git a/drivers/net/ethernet/cavium/liquidio/octeon_console.c b/drivers/net/ethernet/cavium/liquidio/octeon_console.c
index bd6baf2872a5..f1f0d7a0309a 100644
--- a/drivers/net/ethernet/cavium/liquidio/octeon_console.c
+++ b/drivers/net/ethernet/cavium/liquidio/octeon_console.c
@@ -802,19 +802,17 @@ static int octeon_console_read(struct octeon_device *oct, u32 console_num,
}
#define FBUF_SIZE (4 * 1024 * 1024)
-#define MAX_BOOTTIME_SIZE 80
int octeon_download_firmware(struct octeon_device *oct, const u8 *data,
size_t size)
{
struct octeon_firmware_file_header *h;
- char boottime[MAX_BOOTTIME_SIZE];
struct timespec64 ts;
u32 crc32_result;
+ u32 i, rem, used;
u64 load_addr;
u32 image_len;
int ret = 0;
- u32 i, rem;
if (size < sizeof(struct octeon_firmware_file_header)) {
dev_err(&oct->pci_dev->dev, "Firmware file too small (%d < %d).\n",
@@ -896,16 +894,15 @@ int octeon_download_firmware(struct octeon_device *oct, const u8 *data,
* Octeon always uses UTC time. so timezone information is not sent.
*/
ktime_get_real_ts64(&ts);
- ret = snprintf(boottime, MAX_BOOTTIME_SIZE,
+
+ used = strnlen(h->bootcmd, sizeof(h->bootcmd));
+ ret = snprintf(h->bootcmd + used, sizeof(h->bootcmd) - used,
" time_sec=%lld time_nsec=%ld",
(s64)ts.tv_sec, ts.tv_nsec);
- if ((sizeof(h->bootcmd) - strnlen(h->bootcmd, sizeof(h->bootcmd))) <=
- ret) {
+ if (ret >= sizeof(h->bootcmd) - used) {
dev_err(&oct->pci_dev->dev, "Boot command buffer too small\n");
return -EINVAL;
}
- strncat(h->bootcmd, boottime,
- sizeof(h->bootcmd) - strnlen(h->bootcmd, sizeof(h->bootcmd)) - 1);
dev_info(&oct->pci_dev->dev, "Writing boot command: %s\n",
h->bootcmd);
--
2.34.1
In order to append the 'boottime' string to 'h->bootcmd', the final NULL
has to betaken into account when checking if there is enough space.
Fixes: 907aaa6babe1 ("liquidio: pass date and time info to NIC firmware")
Signed-off-by: Christophe JAILLET <[email protected]>
---
drivers/net/ethernet/cavium/liquidio/octeon_console.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/drivers/net/ethernet/cavium/liquidio/octeon_console.c b/drivers/net/ethernet/cavium/liquidio/octeon_console.c
index 67c3570f875f..bd6baf2872a5 100644
--- a/drivers/net/ethernet/cavium/liquidio/octeon_console.c
+++ b/drivers/net/ethernet/cavium/liquidio/octeon_console.c
@@ -899,13 +899,13 @@ int octeon_download_firmware(struct octeon_device *oct, const u8 *data,
ret = snprintf(boottime, MAX_BOOTTIME_SIZE,
" time_sec=%lld time_nsec=%ld",
(s64)ts.tv_sec, ts.tv_nsec);
- if ((sizeof(h->bootcmd) - strnlen(h->bootcmd, sizeof(h->bootcmd))) <
+ if ((sizeof(h->bootcmd) - strnlen(h->bootcmd, sizeof(h->bootcmd))) <=
ret) {
dev_err(&oct->pci_dev->dev, "Boot command buffer too small\n");
return -EINVAL;
}
strncat(h->bootcmd, boottime,
- sizeof(h->bootcmd) - strnlen(h->bootcmd, sizeof(h->bootcmd)));
+ sizeof(h->bootcmd) - strnlen(h->bootcmd, sizeof(h->bootcmd)) - 1);
dev_info(&oct->pci_dev->dev, "Writing boot command: %s\n",
h->bootcmd);
--
2.34.1
On 10/22/2023 1:59 PM, Christophe JAILLET wrote:
> In order to remove the usage of strncat(), write directly at the rigth
> place in the 'h->bootcmd' array and check if the output is truncated.
>
> Signed-off-by: Christophe JAILLET <[email protected]>
> ---
> The goal is to potentially remove the strncat() function from the kernel.
> Their are only few users and most of them use it wrongly.
>
> This patch is compile tested only.
> ---
> .../net/ethernet/cavium/liquidio/octeon_console.c | 13 +++++--------
> 1 file changed, 5 insertions(+), 8 deletions(-)
>
> diff --git a/drivers/net/ethernet/cavium/liquidio/octeon_console.c b/drivers/net/ethernet/cavium/liquidio/octeon_console.c
> index bd6baf2872a5..f1f0d7a0309a 100644
> --- a/drivers/net/ethernet/cavium/liquidio/octeon_console.c
> +++ b/drivers/net/ethernet/cavium/liquidio/octeon_console.c
> @@ -802,19 +802,17 @@ static int octeon_console_read(struct octeon_device *oct, u32 console_num,
> }
>
> #define FBUF_SIZE (4 * 1024 * 1024)
> -#define MAX_BOOTTIME_SIZE 80
>
> int octeon_download_firmware(struct octeon_device *oct, const u8 *data,
> size_t size)
> {
> struct octeon_firmware_file_header *h;
> - char boottime[MAX_BOOTTIME_SIZE];
> struct timespec64 ts;
> u32 crc32_result;
> + u32 i, rem, used;
> u64 load_addr;
> u32 image_len;
> int ret = 0;
> - u32 i, rem;
>
> if (size < sizeof(struct octeon_firmware_file_header)) {
> dev_err(&oct->pci_dev->dev, "Firmware file too small (%d < %d).\n",
> @@ -896,16 +894,15 @@ int octeon_download_firmware(struct octeon_device *oct, const u8 *data,
> * Octeon always uses UTC time. so timezone information is not sent.
> */
> ktime_get_real_ts64(&ts);
> - ret = snprintf(boottime, MAX_BOOTTIME_SIZE,
> +
> + used = strnlen(h->bootcmd, sizeof(h->bootcmd));
> + ret = snprintf(h->bootcmd + used, sizeof(h->bootcmd) - used,
> " time_sec=%lld time_nsec=%ld",
> (s64)ts.tv_sec, ts.tv_nsec);
Maybe I am missing something here, but why not combine the boottime with
the original write of bootcmd? I guess this is being updated
periodically? The overall solution used here feels like the wrong way to
do it... I guess the bootcmd is setup else where but I couldn't find the
relevant code for that.
What you did seems ok to me, but it still feels like there is a simpler
way to achieve the desired result.
Thanks,
Jake
On Sun, 2023-10-22 at 22:59 +0200, Christophe JAILLET wrote:
> In order to remove the usage of strncat(), write directly at the rigth
> place in the 'h->bootcmd' array and check if the output is truncated.
>
> Signed-off-by: Christophe JAILLET <[email protected]>
> ---
> The goal is to potentially remove the strncat() function from the kernel.
> Their are only few users and most of them use it wrongly.
>
> This patch is compile tested only.
Then just switch to strlcat, would be less invasive.
Thanks,
Paolo
On Tue, Oct 24, 2023 at 01:11:13PM +0200, Paolo Abeni wrote:
> On Sun, 2023-10-22 at 22:59 +0200, Christophe JAILLET wrote:
> > In order to remove the usage of strncat(), write directly at the rigth
> > place in the 'h->bootcmd' array and check if the output is truncated.
> >
> > Signed-off-by: Christophe JAILLET <[email protected]>
> > ---
> > The goal is to potentially remove the strncat() function from the kernel.
> > Their are only few users and most of them use it wrongly.
> >
> > This patch is compile tested only.
>
> Then just switch to strlcat, would be less invasive.
Linus was just complaining about strl* functions.
https://lore.kernel.org/all/CAHk-=wj4BZei4JTiX9qsAwk8PEKnPrvkG5FU0i_HNkcDpy7NGQ@mail.gmail.com/
strlcat() does a strlen(src) so it's BROKEN BY DESIGN as Linus puts it.
The advantage of strlcat() is that it always puts a NUL terminator in
the dest buffer, but the disadvantage is that it introduces a read
overflow.
I would probably have written it like this:
diff --git a/drivers/net/ethernet/cavium/liquidio/octeon_console.c b/drivers/net/ethernet/cavium/liquidio/octeon_console.c
index 67c3570f875f..ebe9f7694d8b 100644
--- a/drivers/net/ethernet/cavium/liquidio/octeon_console.c
+++ b/drivers/net/ethernet/cavium/liquidio/octeon_console.c
@@ -899,13 +899,16 @@ int octeon_download_firmware(struct octeon_device *oct, const u8 *data,
ret = snprintf(boottime, MAX_BOOTTIME_SIZE,
" time_sec=%lld time_nsec=%ld",
(s64)ts.tv_sec, ts.tv_nsec);
- if ((sizeof(h->bootcmd) - strnlen(h->bootcmd, sizeof(h->bootcmd))) <
- ret) {
+
+ len = strnlen(h->bootcmd, sizeof(h->bootcmd));
+ len += snprintf(h->bootcmd + len, sizeof(h->bootcmd) - len,
+ " time_sec=%lld time_nsec=%ld",
+ (s64)ts.tv_sec, ts.tv_nsec);
+ if (len >= sizeof(h->bootcmd)) {
+ h->bootcmd[orig] = '\0';
dev_err(&oct->pci_dev->dev, "Boot command buffer too small\n");
return -EINVAL;
}
- strncat(h->bootcmd, boottime,
- sizeof(h->bootcmd) - strnlen(h->bootcmd, sizeof(h->bootcmd)));
dev_info(&oct->pci_dev->dev, "Writing boot command: %s\n",
h->bootcmd);
Don't involve the "ret" variable. Just len +=.
In the original code, if there wasn't enough space they truncated it
before the " time_sec=%lld time_nsec=%ld" but keeping that behavior
seems needlessly complicated. They already created one bug by
complicating stuff.
regards,
dan carpenter