powerpc/powernv: Restrict OPAL symbol map to only be readable by root
Subsystems (2)
PATCH v2 1/1 powerpc/powernv: Restrict OPAL symbol map to only be readable by root
Currently the OPAL symbol map is globally readable, which seems bad as it contains physical addresses.
Restrict it to root.
---
v1->v2:
- fix tabs vs spaces (Greg)
--- arch/powerpc/platforms/powernv/opal.c | 11 +++++++---- 1 file changed, 7 insertions(+), 4 deletions(-)
arch/powerpc/platforms/powernv/opal.c +7 -4
@@ -681,7 +681,10 @@ static ssize_t symbol_map_read(struct file *fp, struct kobject *kobj, bin_attr->size); } -static BIN_ATTR_RO(symbol_map, 0); +static struct bin_attribute symbol_map_attr = { + .attr = {.name ="symbol_map" , .mode =0400 }, + .read = symbol_map_read +};static void opal_export_symmap (void ) {
@@ -698,10 +701,10 @@ static void opal_export_symmap(void)return ;/* Setup attributes */ - bin_attr_symbol_map.private =__va (be64_to_cpu (syms[0 ])); - bin_attr_symbol_map.size =be64_to_cpu (syms[1 ]); + symbol_map_attr.private =__va (be64_to_cpu (syms[0 ])); + symbol_map_attr.size =be64_to_cpu (syms[1 ]); - rc =sysfs_create_bin_file (opal_kobj, &bin_attr_symbol_map); + rc =sysfs_create_bin_file (opal_kobj, &symbol_map_attr);if (rc)pr_warn ("Error %d creating OPAL symbols file\n" , rc); }
Signature · 1 line
2.20.1
Greg KH
maintainerQuoted 26 lines · Andrew Donnellan
Currently the OPAL symbol map is globally readable, which seems bad as it contains physical addresses.
Restrict it to root.
Suggested-by Michael Ellerman <[email protected]>Cc Jordan Niethe <[email protected]>Cc Stewart Smith <[email protected]>Fixes c8742f85125d ("powerpc/powernv: Expose OPAL firmware symbol map")Signed-off-by Andrew Donnellan <[email protected]>---
v1->v2:- fix tabs vs spaces (Greg)
--- arch/powerpc/platforms/powernv/opal.c | 11 +++++++---- 1 file changed, 7 insertions(+), 4 deletions(-)arch/powerpc/platforms/powernv/opal.c +0
-0
arch/powerpc/platforms/powernv/opal.c static ssize_t symbol_map_read(struct file *fp, struct kobject *kobj, · line 687
Quoted 8 lines
arch/powerpc/platforms/powernv/opal.c +4
-1@@ -681,7 +681,10 @@ static ssize_t symbol_map_read(struct file *fp, struct kobject *kobj, bin_attr->size); } -static BIN_ATTR_RO(symbol_map, 0); +static struct bin_attribute symbol_map_attr = { + .attr = {.name ="symbol_map" , .mode =0400 }, + .read = symbol_map_read +};
There's no real need to rename the structure, right? Why not just keep the bin_attr_symbol_map name? That would make this patch even smaller.
arch/powerpc/platforms/powernv/opal.c static void opal_export_symmap(void) · line 707
Quoted 13 lines
arch/powerpc/platforms/powernv/opal.c +3
-3static void opal_export_symmap (void ) {@@ -698,10 +701,10 @@ static void opal_export_symmap(void)return ;/* Setup attributes */ - bin_attr_symbol_map.private =__va (be64_to_cpu (syms[0 ])); - bin_attr_symbol_map.size =be64_to_cpu (syms[1 ]); + symbol_map_attr.private =__va (be64_to_cpu (syms[0 ])); + symbol_map_attr.size =be64_to_cpu (syms[1 ]); - rc =sysfs_create_bin_file (opal_kobj, &bin_attr_symbol_map); + rc =sysfs_create_bin_file (opal_kobj, &symbol_map_attr);
Meta-comment, odds are you are racing userspace when you create this sysfs file, why not add it to the device's default attributes so the driver core creates it for you at the correct time?
thanks,
greg k-h
On 3/5/19 5:59 pm, Greg KH wrote:>> -static BIN_ATTR_RO(symbol_map, 0);
Quoted 7 lines
arch/powerpc/platforms/powernv/opal.c static ssize_t symbol_map_read(struct file *fp, struct kobject *kobj, · line 687
Quoted 4 lines
arch/powerpc/platforms/powernv/opal.c +4
-0+static struct bin_attribute symbol_map_attr = { + .attr = {.name ="symbol_map" , .mode =0400 }, + .read = symbol_map_read +};There's no real need to rename the structure, right? Why not just keep the
bin_attr_symbol_mapname? That would make this patch even smaller.
No real need but it's locally more consistent with the rest of the PPC code. (Though perhaps the other cases should use the BIN_ATTR macro...)
Given this is for stable I'm happy to change that if the smaller patch is more acceptable.
Quoted 12 lines
Quoted 12 lines
static void opal_export_symmap(void) {arch/powerpc/platforms/powernv/opal.c +2
-2@@ -698,10 +701,10 @@ static void opal_export_symmap(void)return ;/* Setup attributes */ - bin_attr_symbol_map.private =__va (be64_to_cpu (syms[0 ])); - bin_attr_symbol_map.size =be64_to_cpu (syms[1 ]); + symbol_map_attr.private =__va (be64_to_cpu (syms[0 ])); + symbol_map_attr.size =be64_to_cpu (syms[1 ]);
Quoted 6 lines
arch/powerpc/platforms/powernv/opal.c static void opal_export_symmap(void) · line 707
Quoted 2 lines
arch/powerpc/platforms/powernv/opal.c +1
-1- rc =sysfs_create_bin_file (opal_kobj, &bin_attr_symbol_map); + rc =sysfs_create_bin_file (opal_kobj, &symbol_map_attr);Meta-comment, odds are you are racing userspace when you create this sysfs file, why not add it to the device's default attributes so the driver core creates it for you at the correct time?
I was not previously aware of default attributes...
Are we actually racing against userspace in a subsys initcall?
Signature · 2 lines
Andrew Donnellan OzLabs, ADL Canberra [email protected] IBM Australia Limited
Greg KH
maintainerQuoted 9 lines · Andrew Donnellan
On 3/5/19 5:59 pm, Greg KH wrote:>> -static
BIN_ATTR_RO(symbol_map, 0);Quoted 7 lines
arch/powerpc/platforms/powernv/opal.c static ssize_t symbol_map_read(struct file *fp, struct kobject *kobj, · line 687
Quoted 4 lines
arch/powerpc/platforms/powernv/opal.c +4
-0+static struct bin_attribute symbol_map_attr = { + .attr = {.name ="symbol_map" , .mode =0400 }, + .read = symbol_map_read +};There's no real need to rename the structure, right? Why not just keep the
bin_attr_symbol_mapname? That would make this patch even smaller.
Quoted 5 lines
No real need but it's locally more consistent with the rest of the PPC code. (Though perhaps the other cases should use the
BIN_ATTRmacro...)Given this is for stable I'm happy to change that if the smaller patch is more acceptable.
stable doesn't care, and if this is more consistent, that's fine with me, I didn't see the larger picture here, just providing unsolicited patch review :)
Quoted 16 lines
Quoted 7 lines
Quoted 7 lines
static void opal_export_symmap(void) {arch/powerpc/platforms/powernv/opal.c +0
-2@@ -698,10 +701,10 @@ static void opal_export_symmap(void)return ;/* Setup attributes */ - bin_attr_symbol_map.private =__va (be64_to_cpu (syms[0 ])); - bin_attr_symbol_map.size =be64_to_cpu (syms[1 ]);Quoted 8 lines
arch/powerpc/platforms/powernv/opal.c static void opal_export_symmap(void) · line 707
Quoted 4 lines
arch/powerpc/platforms/powernv/opal.c +3
-1+ symbol_map_attr.private =__va (be64_to_cpu (syms[0 ])); + symbol_map_attr.size =be64_to_cpu (syms[1 ]); - rc =sysfs_create_bin_file (opal_kobj, &bin_attr_symbol_map); + rc =sysfs_create_bin_file (opal_kobj, &symbol_map_attr);Meta-comment, odds are you are racing userspace when you create this sysfs file, why not add it to the device's default attributes so the driver core creates it for you at the correct time?
Quoted 3 lines
I was not previously aware of default attributes...
Are we actually racing against userspace in a subsys initcall?
You can be, if you subsys is a module :)
thanks,
greg k-h
Quoted 3 lines · Greg KH
Quoted 1 line
Are we actually racing against userspace in a subsys initcall?
You can be, if you subsys is a module :)
For various reasons, we don't compile core system firmware interfaces into modules... that could be an interesting exercise. :D
Signature · 2 lines
Andrew Donnellan OzLabs, ADL Canberra [email protected] IBM Australia Limited
Quoted 11 lines · Andrew Donnellan
Currently the OPAL symbol map is globally readable, which seems bad as it contains physical addresses.
Restrict it to root.
Suggested-by Michael Ellerman <[email protected]>Cc Jordan Niethe <[email protected]>Cc Stewart Smith <[email protected]>Fixes c8742f85125d ("powerpc/powernv: Expose OPAL firmware symbol map")Signed-off-by Andrew Donnellan <[email protected]>
mpe: ping?
Quoted 41 lines
---
v1->v2:- fix tabs vs spaces (Greg)
--- arch/powerpc/platforms/powernv/opal.c | 11 +++++++---- 1 file changed, 7 insertions(+), 4 deletions(-)arch/powerpc/platforms/powernv/opal.c +7
-4@@ -681,7 +681,10 @@ static ssize_t symbol_map_read(struct file *fp, struct kobject *kobj, bin_attr->size); } -static BIN_ATTR_RO(symbol_map, 0); +static struct bin_attribute symbol_map_attr = { + .attr = {.name ="symbol_map" , .mode =0400 }, + .read = symbol_map_read +};static void opal_export_symmap (void ) {@@ -698,10 +701,10 @@ static void opal_export_symmap(void)return ;/* Setup attributes */ - bin_attr_symbol_map.private =__va (be64_to_cpu (syms[0 ])); - bin_attr_symbol_map.size =be64_to_cpu (syms[1 ]); + symbol_map_attr.private =__va (be64_to_cpu (syms[0 ])); + symbol_map_attr.size =be64_to_cpu (syms[1 ]); - rc =sysfs_create_bin_file (opal_kobj, &bin_attr_symbol_map); + rc =sysfs_create_bin_file (opal_kobj, &symbol_map_attr);if (rc)pr_warn ("Error %d creating OPAL symbols file\n" , rc); }
Signature · 2 lines
Andrew Donnellan OzLabs, ADL Canberra [email protected] IBM Australia Limited
Michael Ellerman
reviewerQuoted 6 lines · Andrew Donnellan
Quoted 5 lines · Andrew Donnellan
Currently the OPAL symbol map is globally readable, which seems bad as it contains physical addresses.
Restrict it to root.
Quoted 8 lines
Quoted 6 lines
Suggested-by Michael Ellerman <[email protected]>Cc Jordan Niethe <[email protected]>Cc Stewart Smith <[email protected]>Fixes c8742f85125d ("powerpc/powernv: Expose OPAL firmware symbol map")Signed-off-by Andrew Donnellan <[email protected]>mpe: ping?
Picked up for v5.4.
cheers
Michael Ellerman
AppliedQuoted 11 lines · Andrew Donnellan
Currently the OPAL symbol map is globally readable, which seems bad as it contains physical addresses.
Restrict it to root.
Suggested-by Michael Ellerman <[email protected]>Cc Jordan Niethe <[email protected]>Cc Stewart Smith <[email protected]>Fixes c8742f85125d ("powerpc/powernv: Expose OPAL firmware symbol map")Signed-off-by Andrew Donnellan <[email protected]>
Applied to powerpc next, thanks.
https://git.kernel.org/powerpc/c/e7de4f7b64c23e503a8c42af98d56f2a7462bd6d
cheers