of: reserved_mem: fix OOB write and name skipped nodes
v2 cover of: reserved_mem: fix OOB write and name skipped nodes
Patch 1 fixes an out-of-bounds write in fdt_scan_reserved_mem(). When more than MAX_RESERVED_REGIONS dynamically-placed regions are defined, it writes past the end of the local array.
Patch 2 names the skipped node in fdt_init_reserved_mem_node(), the other place that drops regions.
Changes from v1 [1]
- fix bounds check to cover the missed case in v1, based on sashiko review
- print skipped no name error message in fdt_scan_reserved_mem() and fdt_init_reserved_mem_node() both.
[1] https://lore.kernel.org/all/[email protected]/
Sang-Heon Jeon (2):
of: reserved_mem: prevent OOB when too many dynamic regions are
defined
of: reserved_mem: print skipped node name when too many regions are
defined
drivers/of/of_reserved_mem.c | 17 +++++++++++++---- 1 file changed, 13 insertions(+), 4 deletions(-)
Signature · 1 line
2.43.0
PATCH v2 1/2 of: reserved_mem: prevent OOB when too many dynamic regions are defined
On boot, fdt_scan_reserved_mem() saves each dynamically-placed /reserved-memory subnode into a local array of size MAX_RESERVED_REGIONS.
If the device tree defines more than MAX_RESERVED_REGIONS dynamically-placed regions, fdt_scan_reserved_mem() writes past the end of the local array.
Add a bounds check that logs an error and skips the excess regions, restoring the original behavior.
--- drivers/of/of_reserved_mem.c | 14 +++++++++++--- 1 file changed, 11 insertions(+), 3 deletions(-)
drivers/of/of_reserved_mem.c +11 -3
@@ -359,6 +359,7 @@ int __init fdt_scan_reserved_mem(void) err =__reserved_mem_reserve_reg (child, uname);if (!err) count++; +/* * Save the nodes for the dynamically-placed regions * into an array which will be used for allocation right
@@ -366,10 +367,17 @@ int __init fdt_scan_reserved_mem(void)* or marked as no-map. This is done to avoid dynamically * allocating from one of the statically-placed regions. */ -if (err == -ENOENT &&of_get_flat_dt_prop (child,"size" ,NULL )) { - dynamic_nodes[dynamic_nodes_cnt] = child; - dynamic_nodes_cnt++; +if (err != -ENOENT || !of_get_flat_dt_prop (child,"size" ,NULL )) +continue ; + +if (dynamic_nodes_cnt ==MAX_RESERVED_REGIONS ) { +pr_err ("too many defined dynamic regions, skip '%s'\n" , + uname); +continue ; } + + dynamic_nodes[dynamic_nodes_cnt] = child; + dynamic_nodes_cnt++; }for (int i =0 ; i < dynamic_nodes_cnt; i++) {const char *uname;
Signature · 1 line
2.43.0
[email protected]
AI reviewRe: [PATCH v2 1/2] of: reserved_mem: prevent OOB when too many dynamic regions are defined
Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
Pre-existing issues:
- [Critical] Global buffer overflow on static reserved_mem_array when dynamic allocation fails --
commit ce2daed7217e851888336023e70ee6c49e1f66cd
Author: Sang-Heon Jeon <[email protected]>
of: reserved_mem: prevent OOB when too many dynamic regions are defined
This commit adds bounds checking for dynamically-placed /reserved-memory subnodes to prevent out-of-bounds writes to the local dynamic_nodes array in fdt_scan_reserved_mem().
Quoted 4 lines
drivers/of/of_reserved_mem.c +0
-0
[ ... ]
Quoted 11 lines
drivers/of/of_reserved_mem.c +4
-3@@ -348,10 +349,17 @@ int __init fdt_scan_reserved_mem(void)* or marked as no-map. This is done to avoid dynamically * allocating from one of the statically-placed regions. */ -if (err == -ENOENT &&of_get_flat_dt_prop (child,"size" ,NULL )) { - dynamic_nodes[dynamic_nodes_cnt] = child; - dynamic_nodes_cnt++; +if (err != -ENOENT || !of_get_flat_dt_prop (child,"size" ,NULL )) +continue ; + +if (dynamic_nodes_cnt ==MAX_RESERVED_REGIONS ) {
drivers/of/of_reserved_mem.c int __init fdt_scan_reserved_mem(void) · line 365
Quoted 10 lines
drivers/of/of_reserved_mem.c +6
-0+pr_err ("too many defined dynamic regions, skip '%s'\n" , + uname); +continue ; } + + dynamic_nodes[dynamic_nodes_cnt] = child; + dynamic_nodes_cnt++; }for (int i =0 ; i < dynamic_nodes_cnt; i++) {const char *uname;
[Severity: Critical]
This is a pre-existing issue, but could the initialization sequence still leave the kernel vulnerable to a global buffer overflow later on?
In fdt_scan_reserved_mem(), total_reserved_mem_cnt is updated to the total count of regions, which could exceed MAX_RESERVED_REGIONS due to statically-placed regions.
Later, fdt_scan_reserved_mem_late() calls alloc_reserved_mem_array() to dynamically allocate a larger array:
drivers/of/of_reserved_mem.c:alloc_reserved_mem_array() {
...
new_array = memblock_alloc(alloc_size, SMP_CACHE_BYTES);
if (!new_array) {
pr_err("Failed to allocate memory for reserved_mem array with err: %d", -ENOMEM);
return;
}
...
}If memblock_alloc() fails, it returns early but does not reset total_reserved_mem_cnt back to MAX_RESERVED_REGIONS.
When iterating over the regions, fdt_init_reserved_mem_node() relies on total_reserved_mem_cnt for its bounds checking:
drivers/of/of_reserved_mem.c:fdt_init_reserved_mem_node() {
...
if (reserved_mem_count == total_reserved_mem_cnt) {
pr_err("not enough space for all defined regions, skip '%s'\n",
uname);
return;
}
...
}Since total_reserved_mem_cnt remains inflated, this check will fail to protect the statically-sized reserved_mem_array, leading to an out-of-bounds write.
Should alloc_reserved_mem_array() reset total_reserved_mem_cnt to MAX_RESERVED_REGIONS on the error path to avoid overflowing the static reserved_mem_array?
Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=1
Re: [PATCH v2 1/2] of: reserved_mem: prevent OOB when too many dynamic regions are defined
In reply to Sang-Heon JeonQuoted 8 lines · Sang-Heon Jeon
On boot,
fdt_scan_reserved_mem()saves each dynamically-placed /reserved-memory subnode into a local array of sizeMAX_RESERVED_REGIONS.If the device tree defines more than
MAX_RESERVED_REGIONSdynamically-placed regions,fdt_scan_reserved_mem()writes past the end of the local array.
Quoted 9 lines
Add a bounds check that logs an error and skips the excess regions, restoring the original behavior.
Fixes 8a6e02d0c00e ("of: reserved_mem: Restructure how the reserved memory regions are processed")Signed-off-by Sang-Heon Jeon <[email protected]>--- drivers/of/of_reserved_mem.c | 14 +++++++++++--- 1 file changed, 11 insertions(+), 3 deletions(-)
Applied, thanks!
PATCH v2 2/2 of: reserved_mem: print skipped node name when too many regions are defined
In reply to Sang-Heon JeonWhen too many reserved memory regions are defined, fdt_init_reserved_mem_node() logs an error and skips the region without naming the node.
Print the node name so the skipped node can be identified, matching fdt_scan_reserved_mem().
--- drivers/of/of_reserved_mem.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
drivers/of/of_reserved_mem.c +2 -1
@@ -641,7 +641,8 @@ static void __init fdt_init_reserved_mem_node(unsigned long node, const char *unstruct reserved_mem *rmem = &reserved_mem[reserved_mem_count];if (reserved_mem_count == total_reserved_mem_cnt) { -pr_err ("not enough space for all defined regions.\n" ); +pr_err ("not enough space for all defined regions, skip '%s'\n" , + uname);return ; }
Signature · 1 line
2.43.0
[email protected]
AI reviewRe: [PATCH v2 2/2] of: reserved_mem: print skipped node name when too many regions are defined
Thank you for your contribution! Sashiko AI review found 3 potential issue(s) to consider:
Thank you for your contribution! Sashiko AI review found 3 potential issue(s) to consider:
Pre-existing issues:
- [High] Out-of-bounds write to the static reserved_mem_array if memblock_alloc fails during dynamic array reallocation.
- [High] Ignored memory reservation failures allow unreserved memory to be successfully registered and handed out as exclusive reserved memory.
- [Medium] Dynamically allocated no-map reserved memory regions are leaked from memblock.reserved if their initialization fails. --
commit f95986ab58870e3d79cb320c6dc73ef80982357e
Author: Sang-Heon Jeon <[email protected]>
of: reserved_mem: print skipped node name when too many regions are defined
When too many reserved memory regions are defined, fdt_init_reserved_mem_node() logs an error and skips the region without naming the node.
Print the node name so the skipped node can be identified, matching fdt_scan_reserved_mem().
drivers/of/of_reserved_mem.c static void __init fdt_init_reserved_mem_node(unsigned long node, const char *un · line 629
Quoted 13 lines
drivers/of/of_reserved_mem.c +2
-1@@ -623,7 +623,8 @@ static void __init fdt_init_reserved_mem_node(unsigned long node, const char *unstruct reserved_mem *rmem = &reserved_mem[reserved_mem_count];if (reserved_mem_count == total_reserved_mem_cnt) { -pr_err ("not enough space for all defined regions.\n" ); +pr_err ("not enough space for all defined regions, skip '%s'\n" , + uname);return ; }
[Severity: High]
This isn't a bug introduced by this patch, but does alloc_reserved_mem_array() fail to reset total_reserved_mem_cnt on error?
If memblock_alloc() fails during dynamic array reallocation, it returns early without resetting total_reserved_mem_cnt back to its original value. When the bounds check in fdt_init_reserved_mem_node() is reached later, reserved_mem_count can increment beyond the statically-sized reserved_mem_array limit, potentially causing an out-of-bounds write:
drivers/of/of_reserved_mem.c:alloc_reserved_mem_array() {
...
new_array = memblock_alloc(alloc_size, SMP_CACHE_BYTES);
if (!new_array) {
pr_err("Failed to allocate memory for reserved_mem array...");
return;
}
...
}[Severity: High]
This is a pre-existing issue, but does __reserved_mem_reserve_reg() unconditionally return 0 even if early_init_dt_reserve_memory() fails?
If memory reservation fails, it logs an error but returns success, allowing unreserved memory to be successfully registered and later handed out as exclusive reserved memory:
drivers/of/of_reserved_mem.c:__reserved_mem_reserve_reg() {
...
if (early_init_dt_reserve_memory(base, size, nomap)) {
pr_err("Reserved memory: failed to reserve...");
}
}
return 0;
}[Severity: Medium]
This is a pre-existing issue, but does fdt_init_reserved_mem_node() leak dynamically allocated no-map regions if initialization fails?
If __reserved_mem_init_node() returns an error, the cleanup path for no-map regions clears the MEMBLOCK_NOMAP flag but skips calling memblock_phys_free(), which permanently leaks the physical memory from memblock.reserved:
drivers/of/of_reserved_mem.c:fdt_init_reserved_mem_node() {
...
err = __reserved_mem_init_node(rmem, node);
if (err != 0 && err != -ENODEV) {
pr_info("node %s compatible matching fail\n", rmem->name);
rmem->name = NULL;
if (nomap)
memblock_clear_nomap(rmem->base, rmem->size);
else
memblock_phys_free(rmem->base, rmem->size);
return;
...
}Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=2
Re: [PATCH v2 2/2] of: reserved_mem: print skipped node name when too many regions are defined
Quoted 63 lines · On Sun, Jun 14, 2026 at 10:50 PM <[email protected]> wrote:
Thank you for your contribution! Sashiko AI review found 3 potential issue(s) to consider:
Pre-existing issues:
- [High] Out-of-bounds write to the staticreserved_mem_arrayifmemblock_allocfails during dynamic array reallocation.
- [High] Ignored memory reservation failures allow unreserved memory to be successfully registered and handed out as exclusive reserved memory.
- [Medium] Dynamically allocatedno-mapreserved memory regions are leaked frommemblock.reservedif their initialization fails. --commit
f95986ab58870e3d79cb320c6dc73ef80982357e
Author: Sang-Heon Jeon <[email protected]>of:
reserved_mem: print skipped node name when too many regions are definedWhen too many reserved memory regions are defined,
fdt_init_reserved_mem_node()logs an error and skips the region without naming the node.Print the node name so the skipped node can be identified, matching
fdt_scan_reserved_mem().drivers/of/of_reserved_mem.c static void __init fdt_init_reserved_mem_node(unsigned long node, const char *un · line 629
Quoted 13 lines
drivers/of/of_reserved_mem.c +2
-1@@ -623,7 +623,8 @@ static void __init fdt_init_reserved_mem_node(unsigned long node, const char *unstruct reserved_mem *rmem = &reserved_mem[reserved_mem_count];if (reserved_mem_count == total_reserved_mem_cnt) { -pr_err ("not enough space for all defined regions.\n" ); +pr_err ("not enough space for all defined regions, skip '%s'\n" , + uname);return ; }[Severity: High]
This isn't a bug introduced by this patch, but doesalloc_reserved_mem_array()fail to resettotal_reserved_mem_cnton error?If
memblock_alloc()fails during dynamic array reallocation, it returns early without resettingtotal_reserved_mem_cntback to its original value. When the bounds check infdt_init_reserved_mem_node()is reached later,reserved_mem_countcan increment beyond the statically-sizedreserved_mem_arraylimit, potentially causing an out-of-bounds write:drivers/of/of_reserved_mem.c:alloc_reserved_mem_array() { ... new_array = memblock_alloc(alloc_size, SMP_CACHE_BYTES); if (!new_array) { pr_err("Failed to allocate memory for reserved_mem array..."); return; } ... }[Severity: High]
This is a pre-existing issue, but does__reserved_mem_reserve_reg()unconditionally return 0 even ifearly_init_dt_reserve_memory()fails?If memory reservation fails, it logs an error but returns success, allowing unreserved memory to be successfully registered and later handed out as exclusive reserved memory:
Quoted 8 lines
drivers/of/of_reserved_mem.c:__reserved_mem_reserve_reg() {
...
if (early_init_dt_reserve_memory(base, size, nomap)) {
pr_err("Reserved memory: failed to reserve...");
}
}
return 0;
}These areas are affected by the ongoing series [1], so it might be better not to touch them until it is merged.
[1] https://lore.kernel.org/all/[email protected]/
Quoted 8 lines
[Severity: Medium]
This is a pre-existing issue, but doesfdt_init_reserved_mem_node()leak dynamically allocated no-map regions if initialization fails?If
__reserved_mem_init_node()returns an error, the cleanup path for no-map regions clears theMEMBLOCK_NOMAPflag but skips callingmemblock_phys_free(), which permanently leaks the physical memory from memblock.reserved:
Quoted 14 lines
drivers/of/of_reserved_mem.c:fdt_init_reserved_mem_node() {
...
err = __reserved_mem_init_node(rmem, node);
if (err != 0 && err != -ENODEV) {
pr_info("node %s compatible matching fail\n", rmem->name);
rmem->name = NULL;
if (nomap)
memblock_clear_nomap(rmem->base, rmem->size);
else
memblock_phys_free(rmem->base, rmem->size);
return;
...
}It should be rare, but looks correct to me. BTW, it is not directly related to this patchset, so I think it is better to fix it in a separate patch.
Quoted 2 lines
--
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=2
Best Regards,
Sang-Heon Jeon
Re: [PATCH v2 2/2] of: reserved_mem: print skipped node name when too many regions are defined
In reply to Sang-Heon JeonQuoted 4 lines · Sang-Heon Jeon
When too many reserved memory regions are defined,
fdt_init_reserved_mem_node()logs an error and skips the region without naming the node.
Quoted 8 lines
Print the node name so the skipped node can be identified, matching
fdt_scan_reserved_mem().Signed-off-by Sang-Heon Jeon <[email protected]>--- drivers/of/of_reserved_mem.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
Applied, thanks!