Skip to contentSkip to conversation

efifb: prevent null-deref when iterating dmi_list

PATCH Superseded David Herrmann 3 participants 8 messages Oct 2, 2013 – Nov 1, 2013 Framebuffer layer

lore
Share & tools
lore
#0 lore raw

PATCH v2 1/1 efifb: prevent null-deref when iterating dmi_list

The dmi_list array is initialized using gnu designated initializers, and therefore may contain fewer explicitly defined entries as there are elements in it. This is because the enum above with M_xyz constants contains more items than the designated initializer. Those elements not explicitly initialized are implicitly set to 0.

Now efifb_setup() loops through all these array elements, and performs a strcmp on each item. For non explicitly initialized elements this will be a null pointer:

This patch swaps the check order in the if statement, thus checks first whether dmi_list[i].base is null.

---
Hi

As James didn't respond to the last emails, I just rebased the patch and resent it. The efi M_xyz constants were moved to x86-sysfb so if anyone wants to remove unused bits, please send a separate patch to LKML and x86-ML. This patch just fixes the NULL-deref.

Thanks
David

 drivers/video/efifb.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/video/efifb.c +2 -2
index 7f9ff75..fcb9500 100644
@@ -108,8 +108,8 @@ static int efifb_setup(char *options)
 			if (!*this_opt) continue;
 
 			for (i = 0; i < M_UNKNOWN; i++) {
-				if (!strcmp(this_opt, efifb_dmi_list[i].optname) &&
-				    efifb_dmi_list[i].base != 0) {
+				if (efifb_dmi_list[i].base != 0 &&
+				    !strcmp(this_opt, efifb_dmi_list[i].optname)) {
 					screen_info.lfb_base = efifb_dmi_list[i].base;
 					screen_info.lfb_linelength = efifb_dmi_list[i].stride;
 					screen_info.lfb_width = efifb_dmi_list[i].width;
Signature · 1 line
1.8.4
#1 lore raw

PATCH v3 1/1 efifb: prevent null-deref when iterating dmi_list

From: James Bates <[email protected]>

The dmi_list array is initialized using gnu designated initializers, and therefore may contain fewer explicitly defined entries as there are elements in it. This is because the enum above with M_xyz constants contains more items than the designated initializer. Those elements not explicitly initialized are implicitly set to 0.

Now efifb_setup() loops through all these array elements, and performs a strcmp on each item. For non explicitly initialized elements this will be a null pointer:

This patch swaps the check order in the if statement, thus checks first whether dmi_list[i].base is null.

---
v3: fixes the "Author" field and James' email address

 drivers/video/efifb.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/video/efifb.c +2 -2
index 7f9ff75..fcb9500 100644
@@ -108,8 +108,8 @@ static int efifb_setup(char *options)
 			if (!*this_opt) continue;
 
 			for (i = 0; i < M_UNKNOWN; i++) {
-				if (!strcmp(this_opt, efifb_dmi_list[i].optname) &&
-				    efifb_dmi_list[i].base != 0) {
+				if (efifb_dmi_list[i].base != 0 &&
+				    !strcmp(this_opt, efifb_dmi_list[i].optname)) {
 					screen_info.lfb_base = efifb_dmi_list[i].base;
 					screen_info.lfb_linelength = efifb_dmi_list[i].stride;
 					screen_info.lfb_width = efifb_dmi_list[i].width;
Signature · 1 line
1.8.4
#2 lore raw

ping

Quoted 41 lines · David Herrmann

From: James Bates <[email protected]>

The dmi_list array is initialized using gnu designated initializers, and therefore may contain fewer explicitly defined entries as there are elements in it. This is because the enum above with M_xyz constants contains more items than the designated initializer. Those elements not explicitly initialized are implicitly set to 0.

Now efifb_setup() loops through all these array elements, and performs a strcmp on each item. For non explicitly initialized elements this will be a null pointer:

This patch swaps the check order in the if statement, thus checks first whether dmi_list[i].base is null.

---
v3: fixes the "Author" field and James' email address

 drivers/video/efifb.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/video/efifb.c +2 -3
index 7f9ff75..fcb9500 100644
@@ -108,8 +108,8 @@ static int efifb_setup(char *options)
                        if (!*this_opt) continue;
 
                        for (i = 0; i < M_UNKNOWN; i++) {
-                               if (!strcmp(this_opt, efifb_dmi_list[i].optname) &&
-                                   efifb_dmi_list[i].base != 0) {
+                               if (efifb_dmi_list[i].base != 0 &&
+                                   !strcmp(this_opt, efifb_dmi_list[i].optname)) {
                                        screen_info.lfb_base = efifb_dmi_list[i].base;
                                        screen_info.lfb_linelength = efifb_dmi_list[i].stride;
                                        screen_info.lfb_width = efifb_dmi_list[i].width;
--

1.8.4

#3 lore raw
In reply to David Herrmann
Quoted 29 lines · David Herrmann

From: James Bates <[email protected]>

The dmi_list array is initialized using gnu designated initializers, and therefore may contain fewer explicitly defined entries as there are elements in it. This is because the enum above with M_xyz constants contains more items than the designated initializer. Those elements not explicitly initialized are implicitly set to 0.

Now efifb_setup() loops through all these array elements, and performs a strcmp on each item. For non explicitly initialized elements this will be a null pointer:

This patch swaps the check order in the if statement, thus checks first whether dmi_list[i].base is null.

---
v3: fixes the "Author" field and James' email address

 drivers/video/efifb.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/video/efifb.c +0 -0
index 7f9ff75..fcb9500 100644

drivers/video/efifb.c static int efifb_setup(char *options) · line 115

Quoted 10 lines
drivers/video/efifb.c +2 -2
@@ -108,8 +108,8 @@ static int efifb_setup(char *options)
 			if (!*this_opt) continue;
 
 			for (i = 0; i < M_UNKNOWN; i++) {
-				if (!strcmp(this_opt, efifb_dmi_list[i].optname) &&
-				    efifb_dmi_list[i].base != 0) {
+				if (efifb_dmi_list[i].base != 0 &&
+				    !strcmp(this_opt, efifb_dmi_list[i].optname)) {
 					screen_info.lfb_base = efifb_dmi_list[i].base;
 					screen_info.lfb_linelength = efifb_dmi_list[i].stride;
 					screen_info.lfb_width = efifb_dmi_list[i].width;

Thanks, queued for 3.13.

Tomi

#4 lore raw
In reply to David Herrmann
Quoted 10 lines · David Herrmann

The dmi_list array is initialized using gnu designated initializers, and therefore may contain fewer explicitly defined entries as there are elements in it. This is because the enum above with M_xyz constants contains more items than the designated initializer. Those elements not explicitly initialized are implicitly set to 0.

Now efifb_setup() loops through all these array elements, and performs a strcmp on each item. For non explicitly initialized elements this will be a null pointer:

Quoted 5 lines

This patch swaps the check order in the if statement, thus checks first whether dmi_list[i].base is null.

with the simpleDRM arriving next merge I'm wondering if we need to keep it?

Best Regaards,
J.

Quoted 32 lines

---
Hi

As James didn't respond to the last emails, I just rebased the patch and resent it. The efi M_xyz constants were moved to x86-sysfb so if anyone wants to remove unused bits, please send a separate patch to LKML and x86-ML. This patch just fixes the NULL-deref.

Thanks
David

 drivers/video/efifb.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/video/efifb.c +2 -2
index 7f9ff75..fcb9500 100644
@@ -108,8 +108,8 @@ static int efifb_setup(char *options)
 			if (!*this_opt) continue;
 
 			for (i = 0; i < M_UNKNOWN; i++) {
-				if (!strcmp(this_opt, efifb_dmi_list[i].optname) &&
-				    efifb_dmi_list[i].base != 0) {
+				if (efifb_dmi_list[i].base != 0 &&
+				    !strcmp(this_opt, efifb_dmi_list[i].optname)) {
 					screen_info.lfb_base = efifb_dmi_list[i].base;
 					screen_info.lfb_linelength = efifb_dmi_list[i].stride;
 					screen_info.lfb_width = efifb_dmi_list[i].width;
Signature · 1 line
1.8.4
#5 lore raw

Hi

Quoted 11 lines · Jean-Christophe PLAGNIOL-VILLARD
Quoted 10 lines · David Herrmann

The dmi_list array is initialized using gnu designated initializers, and therefore may contain fewer explicitly defined entries as there are elements in it. This is because the enum above with M_xyz constants contains more items than the designated initializer. Those elements not explicitly initialized are implicitly set to 0.

Now efifb_setup() loops through all these array elements, and performs a strcmp on each item. For non explicitly initialized elements this will be a null pointer:

Quoted 7 lines
Quoted 5 lines

This patch swaps the check order in the if statement, thus checks first whether dmi_list[i].base is null.

with the simpleDRM arriving next merge I'm wondering if we need to keep it?

SimpleDRM is not coming next merge-window. It's basically finished, but I'm still working on the user-space side as its KMS api is highly reduced compared to fully-featured DRM/KMS drivers. Maybe 3.13 will work out.

Anyhow, this patch is still needed as it fixes a serious bug for simplefb.

Thanks
David

#6 lore raw
Quoted 21 lines · David Herrmann

Hi

Quoted 17 lines · Jean-Christophe PLAGNIOL-VILLARD
Quoted 15 lines · David Herrmann

The dmi_list array is initialized using gnu designated initializers, and therefore may contain fewer explicitly defined entries as there are elements in it. This is because the enum above with M_xyz constants contains more items than the designated initializer. Those elements not explicitly initialized are implicitly set to 0.

Now efifb_setup() loops through all these array elements, and performs a strcmp on each item. For non explicitly initialized elements this will be a null pointer:

This patch swaps the check order in the if statement, thus checks first whether dmi_list[i].base is null.

Quoted 6 lines
Quoted 1 line

with the simpleDRM arriving next merge I'm wondering if we need to keep it?

SimpleDRM is not coming next merge-window. It's basically finished, but I'm still working on the user-space side as its KMS api is highly reduced compared to fully-featured DRM/KMS drivers. Maybe 3.13 will work out.

do you have a git tree for the simpleDRM that I can pull?

Quoted 2 lines

Anyhow, this patch is still needed as it fixes a serious bug for simplefb.

ok

Quoted 3 lines

Thanks
David

#7 lore raw

Hi

Quoted 22 lines · Jean-Christophe PLAGNIOL-VILLARD
Quoted 21 lines · David Herrmann

Hi

Quoted 17 lines · Jean-Christophe PLAGNIOL-VILLARD
Quoted 15 lines · David Herrmann

The dmi_list array is initialized using gnu designated initializers, and therefore may contain fewer explicitly defined entries as there are elements in it. This is because the enum above with M_xyz constants contains more items than the designated initializer. Those elements not explicitly initialized are implicitly set to 0.

Now efifb_setup() loops through all these array elements, and performs a strcmp on each item. For non explicitly initialized elements this will be a null pointer:

This patch swaps the check order in the if statement, thus checks first whether dmi_list[i].base is null.

Quoted 8 lines
Quoted 6 lines
Quoted 1 line

with the simpleDRM arriving next merge I'm wondering if we need to keep it?

SimpleDRM is not coming next merge-window. It's basically finished, but I'm still working on the user-space side as its KMS api is highly reduced compared to fully-featured DRM/KMS drivers. Maybe 3.13 will work out.

do you have a git tree for the simpleDRM that I can pull?

Sure, I usually push it to my fdo tree:
http://cgit.freedesktop.org/~dvdhrm/linux/log/?h=simpledrm
But note that you shouldn't be using it right now. User-space fails on it as SimpleDRM only provides a single KMS-FB. It also needs to be adjusted to the new x86-sysfb changes and I am reworking the handover to real drivers.

Cheers
David