| Message ID | 8db90c8f5124bbeab20af9c1314439da8f033184.1627600430.git.gustavoars@kernel.org (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | Fix size comparison bug and use flexible array | expand |
On Thu, Jul 29, 2021 at 06:22:40PM -0500, Gustavo A. R. Silva wrote: > There is a regular need in the kernel to provide a way to declare having > a dynamically sized set of trailing elements in a structure. Kernel code > should always use “flexible array members”[1] for these cases. The older > style of one-element or zero-length arrays should no longer be used[2]. > > Replace a one-element array with a flexible-array member in struct > imgu_fw_header and use the struct_size() helper. > > This also helps with the ongoing efforts to globally enable > -Warray-bounds and get us closer to being able to tighten the > FORTIFY_SOURCE routines on memcpy(). > > [1] https://en.wikipedia.org/wiki/Flexible_array_member > [2] https://www.kernel.org/doc/html/v5.10/process/deprecated.html#zero-length-and-one-element-arrays > > Link: https://github.com/KSPP/linux/issues/79 > Signed-off-by: Gustavo A. R. Silva <gustavoars@kernel.org> > --- > drivers/staging/media/ipu3/ipu3-css-fw.c | 5 ++--- > drivers/staging/media/ipu3/ipu3-css-fw.h | 2 +- > 2 files changed, 3 insertions(+), 4 deletions(-) > > diff --git a/drivers/staging/media/ipu3/ipu3-css-fw.c b/drivers/staging/media/ipu3/ipu3-css-fw.c > index ab021afff954..3b7df1128840 100644 > --- a/drivers/staging/media/ipu3/ipu3-css-fw.c > +++ b/drivers/staging/media/ipu3/ipu3-css-fw.c > @@ -127,9 +127,8 @@ int imgu_css_fw_init(struct imgu_css *css) > if (css->fw->size < sizeof(struct imgu_fw_header) || ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ Originally this was sizeof() the pointer which is clearly wrong. Then patch 1 changed it to force that binary_header[] had at least one element, but now it's changed again to say that binary_header[] can have zero elements. So either patch 1 or patch 2 is wrong. I feel like the probably the correct fix is to just fold these two patches together and say that binary_header[] with zero elements is allowed. But I don't know this code well. regards, dan carpenter
On Fri, Jul 30, 2021 at 10:49:50AM +0300, Dan Carpenter wrote: > On Thu, Jul 29, 2021 at 06:22:40PM -0500, Gustavo A. R. Silva wrote: > > There is a regular need in the kernel to provide a way to declare having > > a dynamically sized set of trailing elements in a structure. Kernel code > > should always use “flexible array members”[1] for these cases. The older > > style of one-element or zero-length arrays should no longer be used[2]. > > > > Replace a one-element array with a flexible-array member in struct > > imgu_fw_header and use the struct_size() helper. > > > > This also helps with the ongoing efforts to globally enable > > -Warray-bounds and get us closer to being able to tighten the > > FORTIFY_SOURCE routines on memcpy(). > > > > [1] https://en.wikipedia.org/wiki/Flexible_array_member > > [2] https://www.kernel.org/doc/html/v5.10/process/deprecated.html#zero-length-and-one-element-arrays > > > > Link: https://github.com/KSPP/linux/issues/79 > > Signed-off-by: Gustavo A. R. Silva <gustavoars@kernel.org> > > --- > > drivers/staging/media/ipu3/ipu3-css-fw.c | 5 ++--- > > drivers/staging/media/ipu3/ipu3-css-fw.h | 2 +- > > 2 files changed, 3 insertions(+), 4 deletions(-) > > > > diff --git a/drivers/staging/media/ipu3/ipu3-css-fw.c b/drivers/staging/media/ipu3/ipu3-css-fw.c > > index ab021afff954..3b7df1128840 100644 > > --- a/drivers/staging/media/ipu3/ipu3-css-fw.c > > +++ b/drivers/staging/media/ipu3/ipu3-css-fw.c > > @@ -127,9 +127,8 @@ int imgu_css_fw_init(struct imgu_css *css) > > if (css->fw->size < sizeof(struct imgu_fw_header) || > ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ > Originally this was sizeof() the pointer which is clearly wrong. Then > patch 1 changed it to force that binary_header[] had at least one > element, but now it's changed again to say that binary_header[] can have > zero elements. So either patch 1 or patch 2 is wrong. > > I feel like the probably the correct fix is to just fold these two > patches together and say that binary_header[] with zero elements is > allowed. But I don't know this code well. Oops... I forgot to tag this one for stable, too. But it would probably be better to fold both of these into a single patch as you suggest. To me these two pieces of code suggest that binary_header[] should not be declared as a one-element array, but a flexible one instead: 130 if (sizeof(struct imgu_fw_bi_file_h) + 131 css->fwp->file_header.binary_nr * sizeof(struct imgu_fw_info) > 132 css->fw->size) 133 goto bad_fw; ... 147 for (i = 0; i < binary_nr; i++) { 148 struct imgu_fw_info *bi = &css->fwp->binary_header[i]; ... Thanks -- Gustavo
Hi Gustavo, On Fri, Jul 30, 2021 at 03:40:09AM -0500, Gustavo A. R. Silva wrote: > On Fri, Jul 30, 2021 at 10:49:50AM +0300, Dan Carpenter wrote: > > On Thu, Jul 29, 2021 at 06:22:40PM -0500, Gustavo A. R. Silva wrote: > > > There is a regular need in the kernel to provide a way to declare having > > > a dynamically sized set of trailing elements in a structure. Kernel code > > > should always use “flexible array members”[1] for these cases. The older > > > style of one-element or zero-length arrays should no longer be used[2]. > > > > > > Replace a one-element array with a flexible-array member in struct > > > imgu_fw_header and use the struct_size() helper. > > > > > > This also helps with the ongoing efforts to globally enable > > > -Warray-bounds and get us closer to being able to tighten the > > > FORTIFY_SOURCE routines on memcpy(). > > > > > > [1] https://en.wikipedia.org/wiki/Flexible_array_member > > > [2] https://www.kernel.org/doc/html/v5.10/process/deprecated.html#zero-length-and-one-element-arrays > > > > > > Link: https://github.com/KSPP/linux/issues/79 > > > Signed-off-by: Gustavo A. R. Silva <gustavoars@kernel.org> > > > --- > > > drivers/staging/media/ipu3/ipu3-css-fw.c | 5 ++--- > > > drivers/staging/media/ipu3/ipu3-css-fw.h | 2 +- > > > 2 files changed, 3 insertions(+), 4 deletions(-) > > > > > > diff --git a/drivers/staging/media/ipu3/ipu3-css-fw.c b/drivers/staging/media/ipu3/ipu3-css-fw.c > > > index ab021afff954..3b7df1128840 100644 > > > --- a/drivers/staging/media/ipu3/ipu3-css-fw.c > > > +++ b/drivers/staging/media/ipu3/ipu3-css-fw.c > > > @@ -127,9 +127,8 @@ int imgu_css_fw_init(struct imgu_css *css) > > > if (css->fw->size < sizeof(struct imgu_fw_header) || > > ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ > > Originally this was sizeof() the pointer which is clearly wrong. Then > > patch 1 changed it to force that binary_header[] had at least one > > element, but now it's changed again to say that binary_header[] can have > > zero elements. So either patch 1 or patch 2 is wrong. > > > > I feel like the probably the correct fix is to just fold these two > > patches together and say that binary_header[] with zero elements is > > allowed. But I don't know this code well. > > Oops... I forgot to tag this one for stable, too. But it would probably > be better to fold both of these into a single patch as you suggest. > > To me these two pieces of code suggest that binary_header[] should not > be declared as a one-element array, but a flexible one instead: I guess if you look at the data structure alone, you're right. But you'll need at least one binary for the firmware to be useful. So if you'd change the struct as in this patch, an additional check for the number of binaries will be needed. I think I'd keep the patches separate, the first one is a bugfix but the second one is just about making the code a little nicer.
diff --git a/drivers/staging/media/ipu3/ipu3-css-fw.c b/drivers/staging/media/ipu3/ipu3-css-fw.c index ab021afff954..3b7df1128840 100644 --- a/drivers/staging/media/ipu3/ipu3-css-fw.c +++ b/drivers/staging/media/ipu3/ipu3-css-fw.c @@ -127,9 +127,8 @@ int imgu_css_fw_init(struct imgu_css *css) if (css->fw->size < sizeof(struct imgu_fw_header) || css->fwp->file_header.h_size != sizeof(struct imgu_fw_bi_file_h)) goto bad_fw; - if (sizeof(struct imgu_fw_bi_file_h) + - css->fwp->file_header.binary_nr * sizeof(struct imgu_fw_info) > - css->fw->size) + if (struct_size(css->fwp, binary_header, + css->fwp->file_header.binary_nr) > css->fw->size) goto bad_fw; dev_info(dev, "loaded firmware version %.64s, %u binaries, %zu bytes\n", diff --git a/drivers/staging/media/ipu3/ipu3-css-fw.h b/drivers/staging/media/ipu3/ipu3-css-fw.h index 3c078f15a295..c0bc57fd678a 100644 --- a/drivers/staging/media/ipu3/ipu3-css-fw.h +++ b/drivers/staging/media/ipu3/ipu3-css-fw.h @@ -171,7 +171,7 @@ struct imgu_fw_bi_file_h { struct imgu_fw_header { struct imgu_fw_bi_file_h file_header; - struct imgu_fw_info binary_header[1]; /* binary_nr items */ + struct imgu_fw_info binary_header[]; /* binary_nr items */ }; /******************* Firmware functions *******************/
There is a regular need in the kernel to provide a way to declare having a dynamically sized set of trailing elements in a structure. Kernel code should always use “flexible array members”[1] for these cases. The older style of one-element or zero-length arrays should no longer be used[2]. Replace a one-element array with a flexible-array member in struct imgu_fw_header and use the struct_size() helper. This also helps with the ongoing efforts to globally enable -Warray-bounds and get us closer to being able to tighten the FORTIFY_SOURCE routines on memcpy(). [1] https://en.wikipedia.org/wiki/Flexible_array_member [2] https://www.kernel.org/doc/html/v5.10/process/deprecated.html#zero-length-and-one-element-arrays Link: https://github.com/KSPP/linux/issues/79 Signed-off-by: Gustavo A. R. Silva <gustavoars@kernel.org> --- drivers/staging/media/ipu3/ipu3-css-fw.c | 5 ++--- drivers/staging/media/ipu3/ipu3-css-fw.h | 2 +- 2 files changed, 3 insertions(+), 4 deletions(-)