drm/amdgpu/discovery: fix OOB read via unchecked die_offset in IP discovery parsing
Subsystems (3)
PATCH 1/1 drm/amdgpu/discovery: fix OOB read via unchecked die_offset in IP discovery parsing
Three call sites in amdgpu_discovery.c dereference firmware-controlled die_offset values without validating them against adev->discovery.size:
amdgpu_discovery_read_harvest_bit_per_ip() line 776-777
amdgpu_discovery_sysfs_init() line 1298-1299
amdgpu_discovery_reg_base_init() line 1524-1525
In all three sites the pattern is:
die_offset = le16_to_cpu(ihdr->die_info[i].die_offset); dhdr = (struct die_header *)(discovery_bin + die_offset);
die_offset is a firmware-controlled u16 (max 65535). The discovery binary is allocated as adev->discovery.size bytes (DISCOVERY_TMR_SIZE = 10240 by default). No bounds check exists between the le16_to_cpu() call and the pointer cast, so a crafted blob with die_offset >= 10240 produces a pointer past the end of the allocation. The subsequent reads of dhdr->die_id and dhdr->num_ips are then slab-out-of-bounds reads.
The ip_offset advancement inside the inner loop also uses struct_size(ip, base_address, ip->num_base_address) where num_base_address is firmware-controlled, enabling unbounded advancement past the allocation on each iteration.
ASAN report (kernel 7.1.0+, QEMU/x86_64, nokaslr, slub_debug=FZPUA):
==================================================================
BUG: KASAN: slab-out-of-bounds in poc_init+0x453/0x1000 [amd_oob_harness] Read of size 2 at addr ffff88800318a801 by task insmod/22 CPU: 0 UID: 0 PID: 22 Comm: insmod Tainted: G O 7.1.0+ #26 PREEMPTLAZY Tainted: [O]=OOT_MODULE Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.17.0-debian-1.17.0-1 04/01/2014 Call Trace: <TASK> dump_stack_lvl+0x2b/0x40 print_report+0x14f/0x4d0 ? wake_up_klogd_work_func+0x70/0x70 ? poc_exit+0xfc0/0xfc0 [amd_oob_harness] kasan_report+0xd4/0x100 ? poc_init+0x453/0x1000 [amd_oob_harness] ? poc_init+0x453/0x1000 [amd_oob_harness] poc_init+0x453/0x1000 [amd_oob_harness] ? poc_exit+0xfc0/0xfc0 [amd_oob_harness] ? poc_exit+0xfc0/0xfc0 [amd_oob_harness] do_one_initcall+0xb0/0x230 ? initcall_blacklisted+0x150/0x150 ? kasan_unpoison+0x40/0x60 do_init_module+0x263/0x810 ? kasan_save_free_info+0x37/0x50 ? free_module+0x300/0x300 ? kfree+0xf1/0x390 load_module+0x3e12/0x51e0 ? sysvec_apic_timer_interrupt+0xa/0x80 ? asm_sysvec_apic_timer_interrupt+0x16/0x20 ? module_frob_arch_sections+0x20/0x20 ? kernel_read_file+0x4d9/0x790 ? kernel_read_file+0x36c/0x790 init_module_from_file+0x136/0x150 ? __do_sys_init_module+0x180/0x180 ? do_sys_openat2+0xeb/0x140 ? fdget+0x64/0x200 __x64_sys_finit_module+0x39f/0x7a0 ? __x64_sys_init_module+0xc0/0xc0 ? __x64_sys_open+0x180/0x180 do_syscall_64+0x56/0x3f0 entry_SYSCALL_64_after_hwframe+0x4b/0x53 RIP: 0033:0x4d1259 Code: ff c3 66 2e 0f 1f 84 00 00 00 00 00 0f 1f 44 00 00 48 89 f8 48 89 f7 48 89 d6 48 89 ca 4d 89 c2 4d 89 c8 4c 8b 4c 24 08 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 c7 c1 e8 ff ff ff f7 d8 64 89 01 48 RSP: 002b:00007fffb68052f8 EFLAGS: 00000246 ORIG_RAX: 0000000000000139 RAX: ffffffffffffffda RBX: 0000000000000000 RCX: 00000000004d1259 RDX: 0000000000000000 RSI: 000000000a529cf0 RDI: 0000000000000003 RBP: 000000000a529cf0 R08: 0000000000000007 R09: 0000000000000000 R10: 0000000000000000 R11: 0000000000000246 R12: 000000000a528dd0 R13: 0000000000000003 R14: 0000000000000000 R15: 0000000000000000 </TASK> The buggy address belongs to the physical page: page: refcount:0 mapcount:0 mapping:0000000000000000 index:0x0 pfn:0x3188 head: order:2 mapcount:0 entire_mapcount:0 nr_pages_mapped:0 pincount:0 flags: 0x4000000000000040(head|zone=1) page_type: f8(unknown) raw: 4000000000000040 0000000000000000 dead000000000122 0000000000000000 raw: 0000000000000000 0000000000000000 00000000f8000000 0000000000000000
head: 4000000000000040 0000000000000000 dead000000000122 0000000000000000
head: 0000000000000000 0000000000000000 00000000f8000000 0000000000000000
head: 4000000000000002 ffffea00000c6201 00000000ffffffff 00000000ffffffff
head: 0000000000000000 0000000000000000 00000000ffffffff 0000000000000000
page dumped because: kasan: bad access detected
Memory state around the buggy address:
ffff88800318a700: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
ffff88800318a780: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
>ffff88800318a800: fe fe fe fe fe fe fe fe fe fe fe fe fe fe fe fe
^
ffff88800318a880: fe fe fe fe fe fe fe fe fe fe fe fe fe fe fe fe
ffff88800318a900: fe fe fe fe fe fe fe fe fe fe fe fe fe fe fe fe
==================================================================
BUG: KASAN: slab-out-of-bounds in poc_init+0x4b3/0x1000 [amd_oob_harness]
Read of size 2 at addr ffff88800318a803 by task insmod/22
CPU: 0 UID: 0 PID: 22 Comm: insmod Tainted: G B O 7.1.0+ #26 PREEMPTLAZY
Tainted: [B]=BAD_PAGE, [O]=OOT_MODULE
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.17.0-debian-1.17.0-1 04/01/2014
Call Trace:
<TASK>
dump_stack_lvl+0x2b/0x40
print_report+0x14f/0x4d0
? add_taint+0x50/0x70
kasan_report+0xd4/0x100
? poc_init+0x4b3/0x1000 [amd_oob_harness]
? poc_init+0x4b3/0x1000 [amd_oob_harness]
poc_init+0x4b3/0x1000 [amd_oob_harness]
? poc_exit+0xfc0/0xfc0 [amd_oob_harness]
? poc_exit+0xfc0/0xfc0 [amd_oob_harness]
do_one_initcall+0xb0/0x230
? initcall_blacklisted+0x150/0x150
? kasan_unpoison+0x40/0x60
do_init_module+0x263/0x810
? kasan_save_free_info+0x37/0x50
? free_module+0x300/0x300
? kfree+0xf1/0x390
load_module+0x3e12/0x51e0
? sysvec_apic_timer_interrupt+0xa/0x80
? asm_sysvec_apic_timer_interrupt+0x16/0x20
? module_frob_arch_sections+0x20/0x20
? kernel_read_file+0x4d9/0x790
? kernel_read_file+0x36c/0x790
init_module_from_file+0x136/0x150
? __do_sys_init_module+0x180/0x180
? do_sys_openat2+0xeb/0x140
? fdget+0x64/0x200
__x64_sys_finit_module+0x39f/0x7a0
? __x64_sys_init_module+0xc0/0xc0
? __x64_sys_open+0x180/0x180
do_syscall_64+0x56/0x3f0
entry_SYSCALL_64_after_hwframe+0x4b/0x53
RIP: 0033:0x4d1259
Code: ff c3 66 2e 0f 1f 84 00 00 00 00 00 0f 1f 44 00 00 48 89 f8 48 89 f7 48 89 d6 48 89 ca 4d 89 c2 4d 89 c8 4c 8b 4c 24 08 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 c7 c1 e8 ff ff ff f7 d8 64 89 01 48
RSP: 002b:00007fffb68052f8 EFLAGS: 00000246 ORIG_RAX: 0000000000000139
RAX: ffffffffffffffda RBX: 0000000000000000 RCX: 00000000007ffff0 RDI: 0000000000000003
RBP: 000000000a529cf0 R08: 0000000000000007 R09: 0000000000000000
R10: 0000000000000000 R11: 0000000000000246 R12: 000000000a528dd0
R13: 0000000000000003 R14: 0000000000000000 R15: 0000000000000000
</TASK>
The buggy address belongs to the physical page:
page: refcount:0 mapcount:0 mapping:0000000000000000 index:0x0 pfn:0x3188
head: order:2 mapcount:0 entire_mapcount:0 nr_pages_mapped:0 pincount:0
flags: 0x4000000000000040(head|zone=1)
page_type: f8(unknown)
raw: 4000000000000040 0000000000000000 dead000000000122 0000000000000000
raw: 0000000000000000 0000000000000000 00000000f8000000 0000000000000000head: 4000000000000040 0000000000000000 dead000000000122 0000000000000000
head: 0000000000000000 0000000000000000 00000000f8000000 0000000000000000
head: 4000000000000002 ffffea00000c6201 00000000ffffffff 00000000ffffffff
head: 0000000000000000 0000000000000000 00000000ffffffff 0000000000000000
page dumped because: kasan: bad access detected
Memory state around the buggy address:
ffff88800318a700: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
ffff88800318a780: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
>ffff88800318a800: fe fe fe fe fe fe fe fe fe fe fe fe fe fe fe fe
^
ffff88800318a880: fe fe fe fe fe fe fe fe fe fe fe fe fe fe fe fe
ffff88800318a900: fe fe fe fe fe fe fe fe fe fe fe fe fe fe fe fe
==================================================================Fix by adding a bounds check on die_offset against adev->discovery.size before the pointer cast in all three sites.
--- drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c | 15 +++++++++++++++ 1 file changed, 15 insertions(+)
drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c +15 -0
@@ -774,6 +774,11 @@ static void amdgpu_discovery_read_harvest_bit_per_ip(struct amdgpu_device *adev,/* scan harvest bit of all IP data structures */ for (i =0 ; i < num_dies; i++) { die_offset =le16_to_cpu (ihdr->die_info[i].die_offset); +if (die_offset +sizeof (*dhdr) > adev->discovery.size) { +dev_err (adev->dev,"invalid die_offset %u in harvest table\n" , + die_offset); +return ; + } dhdr = (struct die_header *)(discovery_bin + die_offset); num_ips =le16_to_cpu (dhdr->num_ips); ip_offset = die_offset +sizeof (*dhdr);
@@ -1296,6 +1301,11 @@ static int amdgpu_discovery_sysfs_recurse(struct amdgpu_device *adev)struct ip_die_entry *ip_die_entry; die_offset =le16_to_cpu (ihdr->die_info[ii].die_offset); +if (die_offset +sizeof (*dhdr) > adev->discovery.size) { +dev_err (adev->dev,"invalid die_offset %u in sysfs init\n" , + die_offset); +return -EINVAL ; + } dhdr = (struct die_header *)(discovery_bin + die_offset); num_ips =le16_to_cpu (dhdr->num_ips); ip_offset = die_offset +sizeof (*dhdr);
@@ -1522,6 +1532,11 @@ static int amdgpu_discovery_reg_base_init(struct amdgpu_device *adev)for (i =0 ; i < num_dies; i++) { die_offset =le16_to_cpu (ihdr->die_info[i].die_offset); +if (die_offset +sizeof (*dhdr) > adev->discovery.size) { +dev_err (adev->dev,"invalid die_offset %u in reg base init\n" , + die_offset); +return -EINVAL ; + } dhdr = (struct die_header *)(discovery_bin + die_offset); num_ips =le16_to_cpu (dhdr->num_ips); ip_offset = die_offset +sizeof (*dhdr);
Signature · 1 line
2.53.0
[email protected]
AI reviewThank 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] The patch identifies but fails to fix out-of-bounds reads caused by unbounded ip_offset advancement.
- [High] Missing bounds check for firmware-controlled num_dies before accessing the die_info array.
- [High] Missing bounds checks for basic header offsets in the discovery binary parser.
--
commit 9d6788c99594a0d1857e6cde0b70c40338a5e49f
Author: Pavitra Jha <[email protected]>
drm/amdgpu/discovery: fix OOB read via unchecked die_offset in IP discovery parsing
This commit adds bounds checking for die_offset against adev->discovery.size in three IP discovery parsing functions. It aims to prevent out-of-bounds reads when processing firmware-controlled die_offset values.
Quoted 4 lines
The
ip_offsetadvancement inside the inner loop also usesstruct_size(ip, base_address, ip->num_base_address)wherenum_base_addressis firmware-controlled, enabling unbounded advancement past the allocation on each iteration.
[Severity: High]
This is a pre-existing issue, but the commit message identifies this unbounded advancement as a root cause, yet the patch doesn't seem to include a bounds check for it in the loop.
Should there be a check against adev->discovery.size for ip_offset inside the inner loop as well, to prevent reading past the end of the allocation when dereferencing the advanced ip_offset?
drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c static void amdgpu_discovery_read_harvest_bit_per_ip(struct amdgpu_device *adev, · line 774
Quoted 6 lines
drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c +0
-0@@ -774,6 +774,11 @@ static void amdgpu_discovery_read_harvest_bit_per_ip(struct amdgpu_device *adev,/* scan harvest bit of all IP data structures */
[Severity: High]
This is a pre-existing issue, but does the construction of the ihdr pointer lack bounds checking?
In amdgpu_discovery_read_harvest_bit_per_ip(), ihdr is constructed using an unchecked offset:
ihdr = (struct ip_discovery_header *)(discovery_bin +
le16_to_cpu(bhdr->table_list[IP_DISCOVERY].offset));
Could an invalid firmware-provided offset cause an out-of-bounds read when ihdr is later dereferenced?
drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c static void amdgpu_discovery_read_harvest_bit_per_ip(struct amdgpu_device *adev, · line 776
Quoted 2 lines
drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c +0
-0for (i =0 ; i < num_dies; i++) { die_offset =le16_to_cpu (ihdr->die_info[i].die_offset);
[Severity: High]
This is a pre-existing issue, but is it possible for num_dies to exceed the fixed array size of 16 for die_info?
Since num_dies is firmware-controlled, could an unchecked value cause an out-of-bounds read when accessing ihdr->die_info[i]?
Quoted 8 lines
drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c +5
-0+if (die_offset +sizeof (*dhdr) > adev->discovery.size) { +dev_err (adev->dev,"invalid die_offset %u in harvest table\n" , + die_offset); +return ; + } dhdr = (struct die_header *)(discovery_bin + die_offset); num_ips =le16_to_cpu (dhdr->num_ips); ip_offset = die_offset +sizeof (*dhdr);
Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=1
PATCH v2 1/1 drm/amdgpu/discovery: validate table offset before IP discovery header cast
Sashiko AI review of the previous fix flagged three remaining gaps in the discovery blob parser, all stemming from the same root cause: the ip_discovery_header pointer itself is constructed from a firmware- controlled offset with no validation before the cast.
ihdr = (struct ip_discovery_header *)(discovery_bin +
le16_to_cpu(bhdr->table_list[IP_DISCOVERY].offset));This offset is a firmware-controlled u16 read directly from the discovery blob's table_list, with no bounds check against adev->discovery.size before being used to construct ihdr. Every subsequent read from ihdr, including num_dies and die_info[], is downstream of this unchecked pointer.
The other two items in that review (unbounded ip_offset advancement via num_base_address, and num_dies exceeding die_info[]'s capacity) were already addressed in the previous fix.
Fix by validating the table offset in amdgpu_discovery_get_table_info(), which is the common path used by all callers except amdgpu_discovery_read_harvest_bit_per_ip() (which reads table_list[IP_DISCOVERY].offset directly rather than going through get_table_info()). Add the equivalent check at that direct access site as well, so all paths that construct an ip_discovery_header pointer from a table offset are covered.
The check validates the offset itself against adev->discovery.size, independent of any specific downstream struct size, since get_table_info() is shared across ten different table types (IP_DISCOVERY, HARVEST_INFO, GC, MALL_INFO, VCN_INFO, NPS_INFO, and others) each with differently-sized table structures.
--- drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c | 14 ++++++++++++++ 1 file changed, 14 insertions(+)
drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c +14 -0
@@ -564,6 +564,12 @@ static int amdgpu_discovery_get_table_info(struct amdgpu_device *adev,return -EINVAL ; } +if (le16_to_cpu ((*info)->offset) >= adev->discovery.size) { +dev_err (adev->dev,"invalid table offset %u for table_id %u\n" , +le16_to_cpu ((*info)->offset), table_id); +return -EINVAL ; + } +return 0 ; }
@@ -766,6 +772,14 @@ static void amdgpu_discovery_read_harvest_bit_per_ip(struct amdgpu_device *adev,int i, j; bhdr = (struct binary_header *)discovery_bin; + +if (le16_to_cpu (bhdr->table_list[IP_DISCOVERY ].offset) >= + adev->discovery.size) { +dev_err (adev->dev,"invalid IP_DISCOVERY table offset %u\n" , +le16_to_cpu (bhdr->table_list[IP_DISCOVERY ].offset)); +return ; + } + ihdr = (struct ip_discovery_header *)(discovery_bin +le16_to_cpu (bhdr->table_list[IP_DISCOVERY ].offset));
Signature · 1 line
2.53.0
[email protected]
AI reviewRe: [PATCH v2] drm/amdgpu/discovery: validate table offset before IP discovery header cast
Thank you for your contribution! Sashiko AI review found 5 potential issue(s) to consider:
Thank you for your contribution! Sashiko AI review found 5 potential issue(s) to consider:
New issues:
- [Low] Misapplied patch hunk causes compilation errors.
Pre-existing issues:
- [High] The intended security fix is completely missing from amdgpu_discovery_get_table_info, leaving out-of-bounds reads unpatched.
- [High] The newly added offset validation only checks the start offset, failing to verify that the structure fits within the buffer.
- [High] Missing validation for downstream table sizes and internal offsets causes out-of-bounds reads during parsing and checksum verification.
- [High] Direct array access for IP_DISCOVERY table offsets breaks for version 2 binaries.
--
--- Patch [1]: [PATCH v2] drm/amdgpu/discovery: validate table offset before IP discovery header cast --- commit 9aac939f5a3fe8366dddb94f326456646c35f228
Author: Pavitra Jha <[email protected]>
drm/amdgpu/discovery: validate table offset before IP discovery header cast
The commit intends to validate the firmware-controlled table offset before casting it to the IP discovery header to prevent out-of-bounds reads. It aims to add offset checks in both the common parsing path and direct access sites.
Quoted 3 lines
Fix by validating the table offset in
amdgpu_discovery_get_table_info(), which is the common path used by all callers exceptamdgpu_discovery_read_harvest_bit_per_ip()(which reads
[Severity: High]
This isn't a bug introduced by this patch, but the commit message states the fix is added to amdgpu_discovery_get_table_info(), while the diff actually applies the validation hunk to amdgpu_discovery_verify_npsinfo() instead.
Since amdgpu_discovery_get_table_info() remains unpatched, doesn't this leave the shared table parsing paths vulnerable to out-of-bounds reads when parsing firmware data?
Quoted 5 lines
drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c +0
-0
drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c static int amdgpu_discovery_verify_npsinfo(struct amdgpu_device *adev, · line 539
Quoted 9 lines
drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c +6
-0@@ -531,6 +531,12 @@ static int amdgpu_discovery_verify_npsinfo(struct amdgpu_device *adev,return -EINVAL ; } +if (le16_to_cpu ((*info)->offset) >= adev->discovery.size) { +dev_err (adev->dev,"invalid table offset %u for table_id %u\n" , +le16_to_cpu ((*info)->offset), table_id); +return -EINVAL ; + } +
[Severity: Low]
Does this compile?
In amdgpu_discovery_verify_npsinfo(), info is a single pointer of type struct table_info *info, making (*info)->offset invalid C syntax. Additionally, table_id is undeclared in this function. It appears this hunk was intended for amdgpu_discovery_get_table_info() but was misapplied here.
[ ... ]
drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c static void amdgpu_discovery_read_harvest_bit_per_ip(struct amdgpu_device *adev, · line 811
Quoted 4 lines
drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c +0
-0@@ -803,6 +809,14 @@ static void amdgpu_discovery_read_harvest_bit_per_ip(struct amdgpu_device *adev,int i, j; bhdr = (struct binary_header *)discovery_bin;
[Severity: High]
This is a pre-existing issue, but does unconditionally casting to struct binary_header * work for version 2 binaries?
Since version 2 binaries (struct binary_header_v2) insert num_tables and padding before the table_list array, wouldn't accessing bhdr->table_list[IP_DISCOVERY].offset read from those fields instead, leading to incorrect offset calculations and out-of-bounds accesses?
drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c static void amdgpu_discovery_read_harvest_bit_per_ip(struct amdgpu_device *adev, · line 777
Quoted 3 lines
drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c +3
-0+ +if (le16_to_cpu (bhdr->table_list[IP_DISCOVERY ].offset) >= + adev->discovery.size) {
[Severity: High]
This is a pre-existing issue, but does this check adequately prevent out-of-bounds reads?
The check only verifies that the start offset is within the buffer bounds. If an offset is placed at the very end of the buffer, wouldn't the subsequent read of ihdr->num_dies access memory beyond the buffer since the structure size is not validated?
drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c static void amdgpu_discovery_read_harvest_bit_per_ip(struct amdgpu_device *adev, · line 785
Quoted 8 lines
drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c +5
-0+dev_err (adev->dev,"invalid IP_DISCOVERY table offset %u\n" , +le16_to_cpu (bhdr->table_list[IP_DISCOVERY ].offset)); +return ; + } + ihdr = (struct ip_discovery_header *)(discovery_bin +le16_to_cpu (bhdr->table_list[IP_DISCOVERY ].offset));
[Severity: High]
This is a pre-existing issue, but are downstream table sizes and internal offsets validated to prevent out-of-bounds reads?
For example, in amdgpu_discovery_read_harvest_bit_per_ip(), ip_offset is declared as a uint16_t:
ip_offset += struct_size(ip, base_address, ip->num_base_address);
Could a struct_size() overflow result in SIZE_MAX being truncated to a smaller value by uint16_t, thereby defeating built-in overflow protections?
Additionally, in amdgpu_discovery_init(), is it possible for the size calculation to underflow?
size = le16_to_cpu(bhdr->binary_size) - offset;
If binary_size is less than offset, this underflow could lead to a large out-of-bounds read in the checksum verification loop.
Similarly, in amdgpu_discovery_table_check(), table_size is read from headers and passed directly to amdgpu_discovery_verify_checksum(). Is there a check to ensure it fits within the buffer?
Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=1
Re: [PATCH v2] drm/amdgpu/discovery: validate table offset before IP discovery header cast
In reply to Pavitra JhaQuoted 33 lines · Pavitra Jha
Sashiko AI review of the previous fix flagged three remaining gaps in the discovery blob parser, all stemming from the same root cause: the
ip_discovery_headerpointer itself is constructed from a firmware- controlled offset with no validation before the cast.ihdr = (struct ip_discovery_header *)(discovery_bin + le16_to_cpu(bhdr->table_list[IP_DISCOVERY].offset));This offset is a firmware-controlled u16 read directly from the discovery blob's
table_list, with no bounds check againstadev->discovery.sizebefore being used to construct ihdr. Every subsequent read from ihdr, includingnum_diesand die_info[], is downstream of this unchecked pointer.The other two items in that review (unbounded
ip_offsetadvancement vianum_base_address, andnum_diesexceeding die_info[]'s capacity) were already addressed in the previous fix.Fix by validating the table offset in
amdgpu_discovery_get_table_info(), which is the common path used by all callers exceptamdgpu_discovery_read_harvest_bit_per_ip()(which reads table_list[IP_DISCOVERY].offset directly rather than going throughget_table_info()). Add the equivalent check at that direct access site as well, so all paths that construct anip_discovery_headerpointer from a table offset are covered.The check validates the offset itself against
adev->discovery.size, independent of any specific downstream struct size, sinceget_table_info()is shared across ten different table types (IP_DISCOVERY,HARVEST_INFO, GC,MALL_INFO,VCN_INFO,NPS_INFO, and others) each with differently-sized table structures.
Quoted 3 lines
Fixes d0c647a6aae2 ("drm/amdgpu/discovery: support new discovery binary header")Signed-off-by Pavitra Jha <[email protected]>
This patch no longer applies to the current drm-next code. I've fixed it up and attached it here. It was a relatively large change so I want to make sure you are ok with it if I keep your authorship.
Thanks!
Alex
Quoted 37 lines
--- drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c | 14 ++++++++++++++ 1 file changed, 14 insertions(+)drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c +14
-1@@ -564,6 +564,12 @@ static int amdgpu_discovery_get_table_info(struct amdgpu_device *adev,return -EINVAL ; } +if (le16_to_cpu ((*info)->offset) >= adev->discovery.size) { +dev_err (adev->dev,"invalid table offset %u for table_id %u\n" , +le16_to_cpu ((*info)->offset), table_id); +return -EINVAL ; + } +return 0 ; }@@ -766,6 +772,14 @@ static void amdgpu_discovery_read_harvest_bit_per_ip(struct amdgpu_device *adev,int i, j; bhdr = (struct binary_header *)discovery_bin; + +if (le16_to_cpu (bhdr->table_list[IP_DISCOVERY ].offset) >= + adev->discovery.size) { +dev_err (adev->dev,"invalid IP_DISCOVERY table offset %u\n" , +le16_to_cpu (bhdr->table_list[IP_DISCOVERY ].offset)); +return ; + } + ihdr = (struct ip_discovery_header *)(discovery_bin +le16_to_cpu (bhdr->table_list[IP_DISCOVERY ].offset)); --
Quoted 2 lines
2.53.0
From 4bebdc2f6911d5e114b6da58276884ded594ca61 Mon Sep 17 00:00:00 2001
From: Pavitra Jha <[email protected]>
Date: Wed, 8 Jul 2026 02:18:34 -0400
Subject: [PATCH] drm/amdgpu/discovery: validate table offset before IP
discovery header cast
Sashiko AI review of the previous fix flagged three remaining gaps in the discovery blob parser, all stemming from the same root cause: the ip_discovery_header pointer itself is constructed from a firmware- controlled offset with no validation before the cast.
ihdr = (struct ip_discovery_header *)(discovery_bin +
le16_to_cpu(bhdr->table_list[IP_DISCOVERY].offset));This offset is a firmware-controlled u16 read directly from the discovery blob's table_list, with no bounds check against adev->discovery.size before being used to construct ihdr. Every subsequent read from ihdr, including num_dies and die_info[], is downstream of this unchecked pointer.
The other two items in that review (unbounded ip_offset advancement via num_base_address, and num_dies exceeding die_info[]'s capacity) were already addressed in the previous fix.
Fix by validating the table offset in amdgpu_discovery_get_table_info(), which is the common path used by all callers except amdgpu_discovery_read_harvest_bit_per_ip() (which reads table_list[IP_DISCOVERY].offset directly rather than going through get_table_info()). Add the equivalent check at that direct access site as well, so all paths that construct an ip_discovery_header pointer from a table offset are covered.
The check validates the offset itself against adev->discovery.size, independent of any specific downstream struct size, since get_table_info() is shared across ten different table types (IP_DISCOVERY, HARVEST_INFO, GC, MALL_INFO, VCN_INFO, NPS_INFO, and others) each with differently-sized table structures.
v2: rebase on latest code (Alex)
--- drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c | 31 ++++++++++++++++--- 1 file changed, 26 insertions(+), 5 deletions(-)
drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c +26 -5
@@ -566,22 +566,33 @@ static const char *amdgpu_discovery_get_fw_name(struct amdgpu_device *adev) }static struct table_info * -amdgpu_discovery_get_table_info_from_bin (uint8_t *discovery_bin, +amdgpu_discovery_get_table_info_from_bin (struct amdgpu_device *adev, +uint8_t *discovery_bin,uint16_t table_id) {struct binary_header *bhdr = (struct binary_header *)discovery_bin;struct binary_header_v2 *bhdrv2; +struct table_info *info;switch (bhdr->version_major) {case 2 : bhdrv2 = (struct binary_header_v2 *)discovery_bin; -return &bhdrv2->table_list[table_id]; + info = &bhdrv2->table_list[table_id]; +break ;case 1 :case 0 : -return &bhdr->table_list[table_id]; + info = &bhdr->table_list[table_id]; +break ;default :return NULL ; } + +if (le16_to_cpu ((info)->offset) >= adev->discovery.size) { +dev_err (adev->dev,"invalid table offset %u for table_id %u\n" , +le16_to_cpu ((info)->offset), table_id); +return NULL ; + } +return info; } static int amdgpu_discovery_get_table_info(struct amdgpu_device *adev,
@@ -591,7 +602,8 @@ static int amdgpu_discovery_get_table_info(struct amdgpu_device *adev,struct binary_header *bhdr = (struct binary_header *)adev->discovery.bin; - *info =amdgpu_discovery_get_table_info_from_bin (adev->discovery.bin, + *info =amdgpu_discovery_get_table_info_from_bin (adev, + adev->discovery.bin, table_id);if (!*info) { dev_err(adev->dev, "Invalid ip discovery table version %d\n",
@@ -803,6 +815,14 @@ static void amdgpu_discovery_read_harvest_bit_per_ip(struct amdgpu_device *adev,int i, j; bhdr = (struct binary_header *)discovery_bin; + +if (le16_to_cpu (bhdr->table_list[IP_DISCOVERY ].offset) >= + adev->discovery.size) { +dev_err (adev->dev,"invalid IP_DISCOVERY table offset %u\n" , +le16_to_cpu (bhdr->table_list[IP_DISCOVERY ].offset)); +return ; + } + ihdr = (struct ip_discovery_header *)(discovery_bin +le16_to_cpu (bhdr->table_list[IP_DISCOVERY ].offset));
@@ -1329,7 +1349,8 @@ static int amdgpu_discovery_sysfs_recurse(struct amdgpu_device *adev,size_t ip_offset;int ii, res; - info =amdgpu_discovery_get_table_info_from_bin (discovery_bin, + info =amdgpu_discovery_get_table_info_from_bin (adev, + discovery_bin,IP_DISCOVERY );if (!info)return -EINVAL ;
Signature · 1 line
2.55.0
Re: [PATCH v2] drm/amdgpu/discovery: validate table offset before IP discovery header cast
Hi Alex,
Yes, absolutely! You can keep my authorship on the patch.
Thank you for fixing it up.
Best,
Pavitra Jha
Quoted 86 lines · Alex Deucher
Quoted 33 lines · Pavitra Jha
Sashiko AI review of the previous fix flagged three remaining gaps in the discovery blob parser, all stemming from the same root cause: the
ip_discovery_headerpointer itself is constructed from a firmware- controlled offset with no validation before the cast.ihdr = (struct ip_discovery_header *)(discovery_bin + le16_to_cpu(bhdr->table_list[IP_DISCOVERY].offset));This offset is a firmware-controlled u16 read directly from the discovery blob's
table_list, with no bounds check againstadev->discovery.sizebefore being used to construct ihdr. Every subsequent read from ihdr, includingnum_diesand die_info[], is downstream of this unchecked pointer.The other two items in that review (unbounded
ip_offsetadvancement vianum_base_address, andnum_diesexceeding die_info[]'s capacity) were already addressed in the previous fix.Fix by validating the table offset in
amdgpu_discovery_get_table_info(), which is the common path used by all callers exceptamdgpu_discovery_read_harvest_bit_per_ip()(which reads table_list[IP_DISCOVERY].offset directly rather than going throughget_table_info()). Add the equivalent check at that direct access site as well, so all paths that construct anip_discovery_headerpointer from a table offset are covered.The check validates the offset itself against
adev->discovery.size, independent of any specific downstream struct size, sinceget_table_info()is shared across ten different table types (IP_DISCOVERY,HARVEST_INFO, GC,MALL_INFO,VCN_INFO,NPS_INFO, and others) each with differently-sized table structures.Quoted 3 lines
Fixes d0c647a6aae2 ("drm/amdgpu/discovery: support new discovery binary header")Signed-off-by Pavitra Jha <[email protected]>This patch no longer applies to the current drm-next code. I've fixed it up and attached it here. It was a relatively large change so I want to make sure you are ok with it if I keep your authorship.
Thanks!
Alex
Quoted 39 lines
--- drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c | 14 ++++++++++++++ 1 file changed, 14 insertions(+)drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c +14
-1@@ -564,6 +564,12 @@ static int amdgpu_discovery_get_table_info(struct amdgpu_device *adev,return -EINVAL ; } +if (le16_to_cpu ((*info)->offset) >= adev->discovery.size) { +dev_err (adev->dev,"invalid table offset %u for table_id %u\n" , +le16_to_cpu ((*info)->offset), table_id); +return -EINVAL ; + } +return 0 ; }@@ -766,6 +772,14 @@ static void amdgpu_discovery_read_harvest_bit_per_ip(struct amdgpu_device *adev,int i, j; bhdr = (struct binary_header *)discovery_bin; + +if (le16_to_cpu (bhdr->table_list[IP_DISCOVERY ].offset) >= + adev->discovery.size) { +dev_err (adev->dev,"invalid IP_DISCOVERY table offset %u\n" , +le16_to_cpu (bhdr->table_list[IP_DISCOVERY ].offset)); +return ; + } + ihdr = (struct ip_discovery_header *)(discovery_bin +le16_to_cpu (bhdr->table_list[IP_DISCOVERY ].offset)); --2.53.0
kernel test robot
build botRe: [PATCH v2] drm/amdgpu/discovery: validate table offset before IP discovery header cast
In reply to Pavitra Jhakernel test robot noticed the following build errors:
Hi Pavitra,
kernel test robot noticed the following build errors:
[auto build test ERROR on drm-misc/drm-misc-next]
[also build test ERROR on linus/master v7.2-rc7 next-20260810] [If your patch is applied to the wrong git tree, kindly drop us a note. And when submitting patch, we suggest to use '--base' as documented in
https://git-scm.com/docs/git-format-patch#_base_tree_information]
url: https://github.com/intel-lab-lkp/linux/commits/Pavitra-Jha/drm-amdgpu-discovery-validate-table-offset-before-IP-discovery-header-cast/20260812-100224 base: https://gitlab.freedesktop.org/drm/misc/kernel.git drm-misc-next patch link: https://lore.kernel.org/r/20260708061835.111986-1-jhapavitra98%40gmail.com patch subject: [PATCH v2] drm/amdgpu/discovery: validate table offset before IP discovery header cast config: alpha-allyesconfig (https://download.01.org/0day-ci/archive/20260812/[email protected]/config) compiler: alpha-linux-gcc (GCC) 16.1.0 reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20260812/[email protected]/reproduce)
If you fix the issue in a separate patch/commit (i.e. not just a new version of the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <[email protected]>
| Closes: https://lore.kernel.org/oe-kbuild-all/[email protected]/
All errors (new ones prefixed by >>):
In file included from include/linux/byteorder/little_endian.h:5,
from arch/alpha/include/uapi/asm/byteorder.h:5,
from include/asm-generic/bitops/le.h:6,
from arch/alpha/include/asm/bitops.h:472,
from include/linux/bitops.h:67,
from include/linux/thread_info.h:27,
from include/asm-generic/preempt.h:5,
from ./arch/alpha/include/generated/asm/preempt.h:1,
from include/linux/preempt.h:79,
from include/linux/spinlock.h:56,
from include/linux/mmzone.h:8,
from include/linux/gfp.h:7,
from include/linux/firmware.h:8,
from drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c:24:
drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c: In function 'amdgpu_discovery_verify_npsinfo':
>> drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c:534:32: error: invalid type argument of '->' (have 'struct table_info')
534 | if (le16_to_cpu((*info)->offset) >= adev->discovery.size) {
| ^~
include/uapi/linux/byteorder/little_endian.h:37:51: note: in definition of macro '__le16_to_cpu'
37 | #define __le16_to_cpu(x) ((__force __u16)(__le16)(x))
| ^
drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c:534:13: note: in expansion of macro 'le16_to_cpu'
534 | if (le16_to_cpu((*info)->offset) >= adev->discovery.size) {
| ^~~~~~~~~~~
drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c:536:44: error: invalid type argument of '->' (have 'struct table_info')
536 | le16_to_cpu((*info)->offset), table_id);
| ^~
include/uapi/linux/byteorder/little_endian.h:37:51: note: in definition of macro '__le16_to_cpu'
37 | #define __le16_to_cpu(x) ((__force __u16)(__le16)(x))
| ^
include/linux/dev_printk.h:154:9: note: in expansion of macro 'dev_printk_index_wrap'
154 | dev_printk_index_wrap(_dev_err, KERN_ERR, dev, dev_fmt(fmt), ##__VA_ARGS__)
| ^~~~~~~~~~~~~~~~~~~~~
drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c:535:17: note: in expansion of macro 'dev_err'
535 | dev_err(adev->dev, "invalid table offset %u for table_id %u\n",
| ^~~~~~~
In file included from include/linux/device.h:15,
from include/drm/drm_print.h:31,
from drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h:29,
from drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.h:30,
from drivers/gpu/drm/amd/amdgpu/amdgpu.h:37,
from drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c:27:
>> drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c:536:55: error: 'table_id' undeclared (first use in this function); did you mean 'table_info'?
536 | le16_to_cpu((*info)->offset), table_id);
| ^~~~~~~~
include/linux/dev_printk.h:110:37: note: in definition of macro 'dev_printk_index_wrap'
110 | _p_func(dev, fmt, ##__VA_ARGS__); \
| ^~~~~~~~~~~
drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c:535:17: note: in expansion of macro 'dev_err'
535 | dev_err(adev->dev, "invalid table offset %u for table_id %u\n",
| ^~~~~~~
drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c:536:55: note: each undeclared identifier is reported only once for each function it appears in
include/linux/dev_printk.h:110:37: note: in definition of macro 'dev_printk_index_wrap'
110 | _p_func(dev, fmt, ##__VA_ARGS__); \
| ^~~~~~~~~~~
drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c:535:17: note: in expansion of macro 'dev_err'
535 | dev_err(adev->dev, "invalid table offset %u for table_id %u\n",
| ^~~~~~~vim +534 drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c
508
509 static int amdgpu_discovery_verify_npsinfo(struct amdgpu_device *adev,
510 struct table_info *info)
511 {
512 uint8_t *discovery_bin = adev->discovery.bin;
513 uint16_t checksum;
514 uint16_t offset;
515
516 offset = le16_to_cpu(info->offset);
517 checksum = le16_to_cpu(info->checksum);
518
519 struct nps_info_header *nhdr =
520 (struct nps_info_header *)(discovery_bin + offset);
521
522 if (le32_to_cpu(nhdr->table_id) != NPS_INFO_TABLE_ID) {
523 dev_dbg(adev->dev, "invalid ip discovery nps info table id\n");
524 return -EINVAL;
525 }
526
527 if (!amdgpu_discovery_verify_checksum(adev, discovery_bin + offset,
528 le32_to_cpu(nhdr->size_bytes),
529 checksum)) {
530 dev_dbg(adev->dev, "invalid nps info data table checksum\n");
531 return -EINVAL;
532 }
533
> 534 if (le16_to_cpu((*info)->offset) >= adev->discovery.size) {
535 dev_err(adev->dev, "invalid table offset %u for table_id %u\n",
> 536 le16_to_cpu((*info)->offset), table_id); 537 return -EINVAL;
538 }
539
540 return 0;
541 }
542
--
0-DAY CI Kernel Test Service
https://github.com/intel/lkp-tests/wiki
kernel test robot
build botRe: [PATCH v2] drm/amdgpu/discovery: validate table offset before IP discovery header cast
In reply to Pavitra Jhakernel test robot noticed the following build errors:
Hi Pavitra,
kernel test robot noticed the following build errors:
[auto build test ERROR on drm-misc/drm-misc-next]
[also build test ERROR on linus/master v7.2-rc7 next-20260811] [If your patch is applied to the wrong git tree, kindly drop us a note. And when submitting patch, we suggest to use '--base' as documented in
https://git-scm.com/docs/git-format-patch#_base_tree_information]
url: https://github.com/intel-lab-lkp/linux/commits/Pavitra-Jha/drm-amdgpu-discovery-validate-table-offset-before-IP-discovery-header-cast/20260812-100224 base: https://gitlab.freedesktop.org/drm/misc/kernel.git drm-misc-next patch link: https://lore.kernel.org/r/20260708061835.111986-1-jhapavitra98%40gmail.com patch subject: [PATCH v2] drm/amdgpu/discovery: validate table offset before IP discovery header cast config: loongarch-defconfig (https://download.01.org/0day-ci/archive/20260813/[email protected]/config) compiler: clang version 24.0.0git (https://github.com/llvm/llvm-project 12df34b8469b8095359de8c249cb1b2753fadeea) reproduce (this is a W=1 build): (https://download.01.org/0day-ci/archive/20260813/[email protected]/reproduce)
If you fix the issue in a separate patch/commit (i.e. not just a new version of the same patch/commit), kindly add following tags
| Reported-by: kernel test robot <[email protected]>
| Closes: https://lore.kernel.org/oe-kbuild-all/[email protected]/
All errors (new ones prefixed by >>):
>> drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c:534:25: error: member reference type 'struct table_info' is not a pointer; did you mean to use '.'?
534 | if (le16_to_cpu((*info)->offset) >= adev->discovery.size) {
| ~~~~~~~^~
| .
include/linux/byteorder/generic.h:91:21: note: expanded from macro 'le16_to_cpu'
91 | #define le16_to_cpu __le16_to_cpu
| ^
include/uapi/linux/byteorder/little_endian.h:37:51: note: expanded from macro '__le16_to_cpu'
37 | #define __le16_to_cpu(x) ((__force __u16)(__le16)(x))
| ^
drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c:536:23: error: member reference type 'struct table_info' is not a pointer; did you mean to use '.'?
536 | le16_to_cpu((*info)->offset), table_id);
| ~~~~~~~^~
| .
include/linux/dev_printk.h:154:65: note: expanded from macro 'dev_err'
154 | dev_printk_index_wrap(_dev_err, KERN_ERR, dev, dev_fmt(fmt), ##__VA_ARGS__)
| ^~~~~~~~~~~
include/linux/dev_printk.h:110:23: note: expanded from macro 'dev_printk_index_wrap'
110 | _p_func(dev, fmt, ##__VA_ARGS__); \
| ^~~~~~~~~~~
include/uapi/linux/byteorder/little_endian.h:37:51: note: expanded from macro '__le16_to_cpu'
37 | #define __le16_to_cpu(x) ((__force __u16)(__le16)(x))
| ^
>> drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c:536:34: error: use of undeclared identifier 'table_id'
536 | le16_to_cpu((*info)->offset), table_id);
| ^~~~~~~~3 errors generated.
vim +534 drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c
508
509 static int amdgpu_discovery_verify_npsinfo(struct amdgpu_device *adev,
510 struct table_info *info)
511 {
512 uint8_t *discovery_bin = adev->discovery.bin;
513 uint16_t checksum;
514 uint16_t offset;
515
516 offset = le16_to_cpu(info->offset);
517 checksum = le16_to_cpu(info->checksum);
518
519 struct nps_info_header *nhdr =
520 (struct nps_info_header *)(discovery_bin + offset);
521
522 if (le32_to_cpu(nhdr->table_id) != NPS_INFO_TABLE_ID) {
523 dev_dbg(adev->dev, "invalid ip discovery nps info table id\n");
524 return -EINVAL;
525 }
526
527 if (!amdgpu_discovery_verify_checksum(adev, discovery_bin + offset,
528 le32_to_cpu(nhdr->size_bytes),
529 checksum)) {
530 dev_dbg(adev->dev, "invalid nps info data table checksum\n");
531 return -EINVAL;
532 }
533
> 534 if (le16_to_cpu((*info)->offset) >= adev->discovery.size) {
535 dev_err(adev->dev, "invalid table offset %u for table_id %u\n",
> 536 le16_to_cpu((*info)->offset), table_id); 537 return -EINVAL;
538 }
539
540 return 0;
541 }
542
--
0-DAY CI Kernel Test Service
https://github.com/intel/lkp-tests/wiki