Skip to contentSkip to conversation

acpi: fix short OEM [Table] ID padding

PATCH In review Igor Mammedov 4 participants 28 messages Jan 12, 2022 – Feb 1, 2022

lore
Share & tools
lore
#0 lore raw

cover acpi: fix short OEM [Table] ID padding

Since 6.0 the commit:
602b458201 ("acpi: Permit OEM ID and OEM table ID fields to be changed")
regressed values of OEM [Table] ID fields in ACPI tables
by padding them with whitespace is a value is shorter then
max possible. That depending on vendor broke OEM [Table] ID patching with SLIC table values and as result licensing of Windows guests.

First reported here https://gitlab.com/qemu-project/qemu/-/issues/707

CC Marian Postevca <[email protected]>
CC Michael S. Tsirkin <[email protected]>
CC Ani Sinha <[email protected]>

Igor Mammedov (4):
tests: acpi: manually pad OEM_ID/OEM_TABLE_ID for test_oem_fields()
test
tests: acpi: whitelist nvdimm's SSDT and FACP.slic expected blobs
acpi: fix OEM ID/OEM Table ID padding
tests: acpi: update expected blobs

 hw/acpi/aml-build.c              |   4 ++--
 tests/data/acpi/pc/SSDT.dimmpxm  | Bin 734 -> 734 bytes
 tests/data/acpi/q35/FACP.slic    | Bin 244 -> 244 bytes
 tests/data/acpi/q35/SSDT.dimmpxm | Bin 734 -> 734 bytes
 tests/data/acpi/virt/SSDT.memhp  | Bin 736 -> 736 bytes
 tests/qtest/bios-tables-test.c   |  15 ++++++---------
 6 files changed, 8 insertions(+), 11 deletions(-)
Signature · 1 line
2.31.1
#1 lore raw

PATCH 1/4 tests: acpi: manually pad OEM_ID/OEM_TABLE_ID for test_oem_fields() test

The next commit will revert OEM fields padding with whitespace to padding with '\0' as it was before [1]. As result test_oem_fields() will fail due to unexpectedly smaller ID sizes read from QEMU ACPI tables.

Pad OEM_ID/OEM_TABLE_ID manually with spaces so that values the test puts on QEMU CLI and expected values match.

1) 602b458201 ("acpi: Permit OEM ID and OEM table ID fields to be changed")

---
 tests/qtest/bios-tables-test.c | 15 ++++++---------
 1 file changed, 6 insertions(+), 9 deletions(-)
tests/qtest/bios-tables-test.c +6 -9
index e6b72d9026..90c9f6a0a2 100644
@@ -71,9 +71,10 @@
 
 #define ACPI_REBUILD_EXPECTED_AML "TEST_ACPI_REBUILD_AML"
 
-#define OEM_ID             "TEST"
-#define OEM_TABLE_ID       "OEM"
-#define OEM_TEST_ARGS      "-machine x-oem-id="OEM_ID",x-oem-table-id="OEM_TABLE_ID
+#define OEM_ID             "TEST  "
+#define OEM_TABLE_ID       "OEM     "
+#define OEM_TEST_ARGS      "-machine x-oem-id='" OEM_ID "',x-oem-table-id='" \
+                           OEM_TABLE_ID "'"
 
 typedef struct {
     bool tcg_only;
@@ -1519,11 +1520,7 @@ static void test_acpi_q35_slic(void)
 static void test_oem_fields(test_data *data)
 {
     int i;
-    char oem_id[6];
-    char oem_table_id[8];
 
-    strpadcpy(oem_id, sizeof oem_id, OEM_ID, ' ');
-    strpadcpy(oem_table_id, sizeof oem_table_id, OEM_TABLE_ID, ' ');
     for (i = 0; i < data->tables->len; ++i) {
         AcpiSdtTable *sdt;
 
@@ -1533,8 +1530,8 @@ static void test_oem_fields(test_data *data)
             continue;
         }
 
-        g_assert(memcmp(sdt->aml + 10, oem_id, 6) == 0);
-        g_assert(memcmp(sdt->aml + 16, oem_table_id, 8) == 0);
+        g_assert(memcmp(sdt->aml + 10, OEM_ID, 6) == 0);
+        g_assert(memcmp(sdt->aml + 16, OEM_TABLE_ID, 8) == 0);
     }
 }
 
Signature · 1 line
2.31.1
#2 lore raw

Re: [PATCH 1/4] tests: acpi: manually pad OEM_ID/OEM_TABLE_ID for test_oem_fields() test

Quoted 9 lines · Igor Mammedov

The next commit will revert OEM fields padding with whitespace to padding with '\0' as it was before [1]. As result test_oem_fields() will fail due to unexpectedly smaller ID sizes read from QEMU ACPI tables.

Pad OEM_ID/OEM_TABLE_ID manually with spaces so that values the test puts on QEMU CLI and expected values match.

1) 602b458201 ("acpi: Permit OEM ID and OEM table ID fields to be changed")

That's kind of ugly in that we do not test shorter names then. How about we pad with \0 instead? And add a comment explaining why it's done.

Quoted 47 lines
---
 tests/qtest/bios-tables-test.c | 15 ++++++---------
 1 file changed, 6 insertions(+), 9 deletions(-)
tests/qtest/bios-tables-test.c +6 -9
index e6b72d9026..90c9f6a0a2 100644
@@ -71,9 +71,10 @@
 
 #define ACPI_REBUILD_EXPECTED_AML "TEST_ACPI_REBUILD_AML"
 
-#define OEM_ID             "TEST"
-#define OEM_TABLE_ID       "OEM"
-#define OEM_TEST_ARGS      "-machine x-oem-id="OEM_ID",x-oem-table-id="OEM_TABLE_ID
+#define OEM_ID             "TEST  "
+#define OEM_TABLE_ID       "OEM     "
+#define OEM_TEST_ARGS      "-machine x-oem-id='" OEM_ID "',x-oem-table-id='" \
+                           OEM_TABLE_ID "'"
 
 typedef struct {
     bool tcg_only;
@@ -1519,11 +1520,7 @@ static void test_acpi_q35_slic(void)
 static void test_oem_fields(test_data *data)
 {
     int i;
-    char oem_id[6];
-    char oem_table_id[8];
 
-    strpadcpy(oem_id, sizeof oem_id, OEM_ID, ' ');
-    strpadcpy(oem_table_id, sizeof oem_table_id, OEM_TABLE_ID, ' ');
     for (i = 0; i < data->tables->len; ++i) {
         AcpiSdtTable *sdt;
 
@@ -1533,8 +1530,8 @@ static void test_oem_fields(test_data *data)
             continue;
         }
 
-        g_assert(memcmp(sdt->aml + 10, oem_id, 6) == 0);
-        g_assert(memcmp(sdt->aml + 16, oem_table_id, 8) == 0);
+        g_assert(memcmp(sdt->aml + 10, OEM_ID, 6) == 0);
+        g_assert(memcmp(sdt->aml + 16, OEM_TABLE_ID, 8) == 0);
     }
 }
 
Signature · 1 line
2.31.1
#3 lore raw

Re: [PATCH 1/4] tests: acpi: manually pad OEM_ID/OEM_TABLE_ID for test_oem_fields() test

Quoted 5 lines · On Wed, 12 Jan 2022 08:44:19 -0500 "Michael S. Tsirkin" <[email protected]> wrote:
Quoted 4 lines · Igor Mammedov

The next commit will revert OEM fields padding with whitespace to padding with '\0' as it was before [1]. As result test_oem_fields() will fail due to unexpectedly smaller ID sizes read from QEMU ACPI tables.

Quoted 8 lines
Quoted 5 lines

Pad OEM_ID/OEM_TABLE_ID manually with spaces so that values the test puts on QEMU CLI and expected values match.

1) 602b458201 ("acpi: Permit OEM ID and OEM table ID fields to be changed")

That's kind of ugly in that we do not test shorter names then. How about we pad with \0 instead?

test_acpi_q35_slic() should cover short OEM_TABLE_ID.

also padding in this patch makes test_oem_fields() cleaner and simplifies 3/4, switching to \0 here would require merging this patch with the fix itself to avoid breaking bisection.

If you still prefer to have test_oem_fields() test short names, I can post following on top that can to it without breaking bisection:

tests/qtest/bios-tables-test.c +4 -4
index 90c9f6a0a2..0fd7cf1f89 100644
@@ -71,8 +71,8 @@
 
 #define ACPI_REBUILD_EXPECTED_AML "TEST_ACPI_REBUILD_AML"
 
-#define OEM_ID             "TEST  "
-#define OEM_TABLE_ID       "OEM     "
+#define OEM_ID             "TEST"
+#define OEM_TABLE_ID       "OEM"
 #define OEM_TEST_ARGS      "-machine x-oem-id='" OEM_ID "',x-oem-table-id='" \
                            OEM_TABLE_ID "'"
 
@@ -1530,8 +1530,8 @@ static void test_oem_fields(test_data *data)
             continue;
         }
 
-        g_assert(memcmp(sdt->aml + 10, OEM_ID, 6) == 0);
-        g_assert(memcmp(sdt->aml + 16, OEM_TABLE_ID, 8) == 0);
+        g_assert(strncmp((char *)sdt->aml + 10, OEM_ID, 6) == 0);
+        g_assert(strncmp((char *)sdt->aml + 16, OEM_TABLE_ID, 8) == 0);
     }
 }
 
Quoted 50 lines

And add a comment explaining why it's done.

Quoted 47 lines
---
 tests/qtest/bios-tables-test.c | 15 ++++++---------
 1 file changed, 6 insertions(+), 9 deletions(-)
tests/qtest/bios-tables-test.c +6 -9
index e6b72d9026..90c9f6a0a2 100644
@@ -71,9 +71,10 @@
 
 #define ACPI_REBUILD_EXPECTED_AML "TEST_ACPI_REBUILD_AML"
 
-#define OEM_ID             "TEST"
-#define OEM_TABLE_ID       "OEM"
-#define OEM_TEST_ARGS      "-machine x-oem-id="OEM_ID",x-oem-table-id="OEM_TABLE_ID
+#define OEM_ID             "TEST  "
+#define OEM_TABLE_ID       "OEM     "
+#define OEM_TEST_ARGS      "-machine x-oem-id='" OEM_ID "',x-oem-table-id='" \
+                           OEM_TABLE_ID "'"
 
 typedef struct {
     bool tcg_only;
@@ -1519,11 +1520,7 @@ static void test_acpi_q35_slic(void)
 static void test_oem_fields(test_data *data)
 {
     int i;
-    char oem_id[6];
-    char oem_table_id[8];
 
-    strpadcpy(oem_id, sizeof oem_id, OEM_ID, ' ');
-    strpadcpy(oem_table_id, sizeof oem_table_id, OEM_TABLE_ID, ' ');
     for (i = 0; i < data->tables->len; ++i) {
         AcpiSdtTable *sdt;
 
@@ -1533,8 +1530,8 @@ static void test_oem_fields(test_data *data)
             continue;
         }
 
-        g_assert(memcmp(sdt->aml + 10, oem_id, 6) == 0);
-        g_assert(memcmp(sdt->aml + 16, oem_table_id, 8) == 0);
+        g_assert(memcmp(sdt->aml + 10, OEM_ID, 6) == 0);
+        g_assert(memcmp(sdt->aml + 16, OEM_TABLE_ID, 8) == 0);
     }
 }
 
Signature · 1 line
2.31.1  
#4 lore raw

Re: [PATCH 1/4] tests: acpi: manually pad OEM_ID/OEM_TABLE_ID for test_oem_fields() test

Quoted 34 lines · Igor Mammedov
Quoted 5 lines · On Wed, 12 Jan 2022 08:44:19 -0500 "Michael S. Tsirkin" <[email protected]> wrote:
Quoted 4 lines · Igor Mammedov

The next commit will revert OEM fields padding with whitespace to padding with '\0' as it was before [1]. As result test_oem_fields() will fail due to unexpectedly smaller ID sizes read from QEMU ACPI tables.

Quoted 8 lines
Quoted 5 lines

Pad OEM_ID/OEM_TABLE_ID manually with spaces so that values the test puts on QEMU CLI and expected values match.

1) 602b458201 ("acpi: Permit OEM ID and OEM table ID fields to be changed")

That's kind of ugly in that we do not test shorter names then. How about we pad with \0 instead?

test_acpi_q35_slic() should cover short OEM_TABLE_ID.

also padding in this patch makes test_oem_fields() cleaner and simplifies 3/4, switching to \0 here would require merging this patch with the fix itself to avoid breaking bisection.

If you still prefer to have test_oem_fields() test short names, I can post following on top that can to it without breaking bisection:

tests/qtest/bios-tables-test.c +0 -0
index 90c9f6a0a2..0fd7cf1f89 100644

tests/qtest/bios-tables-test.c line 77

Quoted 9 lines
tests/qtest/bios-tables-test.c +2 -2
@@ -71,8 +71,8 @@
 
 #define ACPI_REBUILD_EXPECTED_AML "TEST_ACPI_REBUILD_AML"
 
-#define OEM_ID             "TEST  "
-#define OEM_TABLE_ID       "OEM     "
+#define OEM_ID             "TEST"
+#define OEM_TABLE_ID       "OEM"
 #define OEM_TEST_ARGS      "-machine x-oem-id='" OEM_ID "',x-oem-table-id='" \
                            OEM_TABLE_ID "'"

Don't we want to revert ARGS change too then?

tests/qtest/bios-tables-test.c static void test_oem_fields(test_data *data) · line 1537

Quoted 11 lines
tests/qtest/bios-tables-test.c +2 -2
@@ -1530,8 +1530,8 @@ static void test_oem_fields(test_data *data)
             continue;
         }
 
-        g_assert(memcmp(sdt->aml + 10, OEM_ID, 6) == 0);
-        g_assert(memcmp(sdt->aml + 16, OEM_TABLE_ID, 8) == 0);
+        g_assert(strncmp((char *)sdt->aml + 10, OEM_ID, 6) == 0);
+        g_assert(strncmp((char *)sdt->aml + 16, OEM_TABLE_ID, 8) == 0);
     }
 }
 

Looks much cleaner to me. OK as a patch on top.

Quoted 51 lines
Quoted 51 lines

And add a comment explaining why it's done.

Quoted 47 lines
---
 tests/qtest/bios-tables-test.c | 15 ++++++---------
 1 file changed, 6 insertions(+), 9 deletions(-)
tests/qtest/bios-tables-test.c +6 -9
index e6b72d9026..90c9f6a0a2 100644
@@ -71,9 +71,10 @@
 
 #define ACPI_REBUILD_EXPECTED_AML "TEST_ACPI_REBUILD_AML"
 
-#define OEM_ID             "TEST"
-#define OEM_TABLE_ID       "OEM"
-#define OEM_TEST_ARGS      "-machine x-oem-id="OEM_ID",x-oem-table-id="OEM_TABLE_ID
+#define OEM_ID             "TEST  "
+#define OEM_TABLE_ID       "OEM     "
+#define OEM_TEST_ARGS      "-machine x-oem-id='" OEM_ID "',x-oem-table-id='" \
+                           OEM_TABLE_ID "'"
 
 typedef struct {
     bool tcg_only;
@@ -1519,11 +1520,7 @@ static void test_acpi_q35_slic(void)
 static void test_oem_fields(test_data *data)
 {
     int i;
-    char oem_id[6];
-    char oem_table_id[8];
 
-    strpadcpy(oem_id, sizeof oem_id, OEM_ID, ' ');
-    strpadcpy(oem_table_id, sizeof oem_table_id, OEM_TABLE_ID, ' ');
     for (i = 0; i < data->tables->len; ++i) {
         AcpiSdtTable *sdt;
 
@@ -1533,8 +1530,8 @@ static void test_oem_fields(test_data *data)
             continue;
         }
 
-        g_assert(memcmp(sdt->aml + 10, oem_id, 6) == 0);
-        g_assert(memcmp(sdt->aml + 16, oem_table_id, 8) == 0);
+        g_assert(memcmp(sdt->aml + 10, OEM_ID, 6) == 0);
+        g_assert(memcmp(sdt->aml + 16, OEM_TABLE_ID, 8) == 0);
     }
 }
 
Signature · 1 line
2.31.1  
#5 lore raw

PATCH 2/4 tests: acpi: whitelist nvdimm's SSDT and FACP.slic expected blobs

In reply to Igor Mammedov

The next commit will revert OEM fields whitespace padding to padding with '\0' as it was before [1]. That will change OEM Table ID for:
* SSDT.*: where it was padded from 6 characters to 8
* FACP.slic: where it was padded from 2 characters to 8
after reverting whitespace padding, it will be replaced with '\0' which effectively will shorten OEM table ID to 6 and 2 characters.

Whitelist affected tables before introducing the change.

1) 602b458201 ("acpi: Permit OEM ID and OEM table ID fields to be changed")

---
 tests/qtest/bios-tables-test-allowed-diff.h | 4 ++++
 1 file changed, 4 insertions(+)
tests/qtest/bios-tables-test-allowed-diff.h +4 -0
index dfb8523c8b..7faa8f53be 100644
@@ -1 +1,5 @@
 /* List of comma-separated changed AML files to ignore */
+"tests/data/acpi/virt/SSDT.memhp",
+"tests/data/acpi/pc/SSDT.dimmpxm",
+"tests/data/acpi/q35/SSDT.dimmpxm",
+"tests/data/acpi/q35/FACP.slic",
Signature · 1 line
2.31.1
#6 lore raw

PATCH 3/4 acpi: fix OEM ID/OEM Table ID padding

In reply to Igor Mammedov

Commit [2] broke original '\0' padding of OEM ID and OEM Table ID fields in headers of ACPI tables. While it doesn't have impact on default values since QEMU uses 6 and 8 characters long values respectively, it broke usecase where IDs are provided on QEMU CLI. It shouldn't affect guest (but may cause licensing verification issues in guest OS).
One of the broken usecases is user supplied SLIC table with IDs shorter than max possible length, where [2] mangles IDs with extra spaces in RSDT and FADT tables whereas guest OS expects those to mirror the respective values of the used SLIC table.

Fix it by replacing whitespace padding with '\0' padding in accordance with [1] and expectations of guest OS

1) ACPI spec, v2.0b

       17.2 AML Grammar Definition
       ...
       //OEM ID of up to 6 characters. If the OEM ID is
       //shorter than 6 characters, it can be terminated
       //with a NULL character.

2)

Fixes 602b458201 ("acpi: Permit OEM ID and OEM table ID fields to be changed")

Resolves: https://gitlab.com/qemu-project/qemu/-/issues/707

---
 hw/acpi/aml-build.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
hw/acpi/aml-build.c +2 -2
index b3b3310df3..65148d5b9d 100644
@@ -1724,9 +1724,9 @@ void acpi_table_begin(AcpiTable *desc, GArray *array)
     build_append_int_noprefix(array, 0, 4); /* Length */
     build_append_int_noprefix(array, desc->rev, 1); /* Revision */
     build_append_int_noprefix(array, 0, 1); /* Checksum */
-    build_append_padded_str(array, desc->oem_id, 6, ' '); /* OEMID */
+    build_append_padded_str(array, desc->oem_id, 6, '\0'); /* OEMID */
     /* OEM Table ID */
-    build_append_padded_str(array, desc->oem_table_id, 8, ' ');
+    build_append_padded_str(array, desc->oem_table_id, 8, '\0');
     build_append_int_noprefix(array, 1, 4); /* OEM Revision */
     g_array_append_vals(array, ACPI_BUILD_APPNAME8, 4); /* Creator ID */
     build_append_int_noprefix(array, 1, 4); /* Creator Revision */
Signature · 1 line
2.31.1
#7 lore raw

Re: [PATCH 3/4] acpi: fix OEM ID/OEM Table ID padding

Quoted 15 lines · Igor Mammedov

Commit [2] broke original '\0' padding of OEM ID and OEM Table ID fields in headers of ACPI tables. While it doesn't have impact on default values since QEMU uses 6 and 8 characters long values respectively, it broke usecase where IDs are provided on QEMU CLI. It shouldn't affect guest (but may cause licensing verification issues in guest OS).
One of the broken usecases is user supplied SLIC table with IDs shorter than max possible length, where [2] mangles IDs with extra spaces in RSDT and FADT tables whereas guest OS expects those to mirror the respective values of the used SLIC table.

Fix it by replacing whitespace padding with '\0' padding in accordance with [1] and expectations of guest OS

1) ACPI spec, v2.0b

Quoted 8 lines
       17.2 AML Grammar Definition
       ...
       //OEM ID of up to 6 characters. If the OEM ID is
       //shorter than 6 characters, it can be terminated
       //with a NULL character.

2)

Fixes 602b458201 ("acpi: Permit OEM ID and OEM table ID fields to be changed")

and add:

?

Quoted 26 lines

Resolves: https://gitlab.com/qemu-project/qemu/-/issues/707

---
 hw/acpi/aml-build.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
hw/acpi/aml-build.c +2 -2
index b3b3310df3..65148d5b9d 100644
@@ -1724,9 +1724,9 @@ void acpi_table_begin(AcpiTable *desc, GArray *array)
     build_append_int_noprefix(array, 0, 4); /* Length */
     build_append_int_noprefix(array, desc->rev, 1); /* Revision */
     build_append_int_noprefix(array, 0, 1); /* Checksum */
-    build_append_padded_str(array, desc->oem_id, 6, ' '); /* OEMID */
+    build_append_padded_str(array, desc->oem_id, 6, '\0'); /* OEMID */
     /* OEM Table ID */
-    build_append_padded_str(array, desc->oem_table_id, 8, ' ');
+    build_append_padded_str(array, desc->oem_table_id, 8, '\0');
     build_append_int_noprefix(array, 1, 4); /* OEM Revision */
     g_array_append_vals(array, ACPI_BUILD_APPNAME8, 4); /* Creator ID */
     build_append_int_noprefix(array, 1, 4); /* Creator Revision */
Signature · 1 line
2.31.1
#8 lore raw

Re: [PATCH 3/4] acpi: fix OEM ID/OEM Table ID padding

Quoted 22 lines · Michael S. Tsirkin
Quoted 21 lines · Igor Mammedov

Commit [2] broke original '\0' padding of OEM ID and OEM Table ID fields in headers of ACPI tables. While it doesn't have impact on default values since QEMU uses 6 and 8 characters long values respectively, it broke usecase where IDs are provided on QEMU CLI. It shouldn't affect guest (but may cause licensing verification issues in guest OS).
One of the broken usecases is user supplied SLIC table with IDs shorter than max possible length, where [2] mangles IDs with extra spaces in RSDT and FADT tables whereas guest OS expects those to mirror the respective values of the used SLIC table.

Fix it by replacing whitespace padding with '\0' padding in accordance with [1] and expectations of guest OS

1) ACPI spec, v2.0b

       17.2 AML Grammar Definition
       ...
       //OEM ID of up to 6 characters. If the OEM ID is
       //shorter than 6 characters, it can be terminated
       //with a NULL character.
Quoted 7 lines
Quoted 2 lines

2)

Fixes 602b458201 ("acpi: Permit OEM ID and OEM table ID fields to be changed")

and add:

He did that already with the "Resolves:" line below.

Quoted 31 lines

?

Quoted 26 lines

Resolves: https://gitlab.com/qemu-project/qemu/-/issues/707

---
 hw/acpi/aml-build.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
hw/acpi/aml-build.c +2 -3
index b3b3310df3..65148d5b9d 100644
@@ -1724,9 +1724,9 @@ void acpi_table_begin(AcpiTable *desc, GArray *array)
     build_append_int_noprefix(array, 0, 4); /* Length */
     build_append_int_noprefix(array, desc->rev, 1); /* Revision */
     build_append_int_noprefix(array, 0, 1); /* Checksum */
-    build_append_padded_str(array, desc->oem_id, 6, ' '); /* OEMID */
+    build_append_padded_str(array, desc->oem_id, 6, '\0'); /* OEMID */
     /* OEM Table ID */
-    build_append_padded_str(array, desc->oem_table_id, 8, ' ');
+    build_append_padded_str(array, desc->oem_table_id, 8, '\0');
     build_append_int_noprefix(array, 1, 4); /* OEM Revision */
     g_array_append_vals(array, ACPI_BUILD_APPNAME8, 4); /* Creator ID */
     build_append_int_noprefix(array, 1, 4); /* Creator Revision */
--

2.31.1

#9 lore raw

Re: [PATCH 3/4] acpi: fix OEM ID/OEM Table ID padding

Quoted 30 lines · Ani Sinha
Quoted 26 lines · Michael S. Tsirkin
Quoted 23 lines · Igor Mammedov

Commit [2] broke original '\0' padding of OEM ID and OEM Table ID fields in headers of ACPI tables. While it doesn't have impact on default values since QEMU uses 6 and 8 characters long values respectively, it broke usecase where IDs are provided on QEMU CLI. It shouldn't affect guest (but may cause licensing verification issues in guest OS).
One of the broken usecases is user supplied SLIC table with IDs shorter than max possible length, where [2] mangles IDs with extra spaces in RSDT and FADT tables whereas guest OS expects those to mirror the respective values of the used SLIC table.

Fix it by replacing whitespace padding with '\0' padding in accordance with [1] and expectations of guest OS

1) ACPI spec, v2.0b

       17.2 AML Grammar Definition
       ...
       //OEM ID of up to 6 characters. If the OEM ID is
       //shorter than 6 characters, it can be terminated
       //with a NULL character.

2)

Fixes 602b458201 ("acpi: Permit OEM ID and OEM table ID fields to be changed")
Quoted 5 lines
Quoted 3 lines

and add:

He did that already with the "Resolves:" line below.

oh sorry

Quoted 31 lines
Quoted 31 lines

?

Quoted 26 lines

Resolves: https://gitlab.com/qemu-project/qemu/-/issues/707

---
 hw/acpi/aml-build.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
hw/acpi/aml-build.c +2 -3
index b3b3310df3..65148d5b9d 100644
@@ -1724,9 +1724,9 @@ void acpi_table_begin(AcpiTable *desc, GArray *array)
     build_append_int_noprefix(array, 0, 4); /* Length */
     build_append_int_noprefix(array, desc->rev, 1); /* Revision */
     build_append_int_noprefix(array, 0, 1); /* Checksum */
-    build_append_padded_str(array, desc->oem_id, 6, ' '); /* OEMID */
+    build_append_padded_str(array, desc->oem_id, 6, '\0'); /* OEMID */
     /* OEM Table ID */
-    build_append_padded_str(array, desc->oem_table_id, 8, ' ');
+    build_append_padded_str(array, desc->oem_table_id, 8, '\0');
     build_append_int_noprefix(array, 1, 4); /* OEM Revision */
     g_array_append_vals(array, ACPI_BUILD_APPNAME8, 4); /* Creator ID */
     build_append_int_noprefix(array, 1, 4); /* Creator Revision */
--

2.31.1

Reviewed-by on v1 3/4
#10 lore raw

Re: [PATCH 3/4] acpi: fix OEM ID/OEM Table ID padding

In reply to Igor Mammedov
Quoted 21 lines · Igor Mammedov

Commit [2] broke original '\0' padding of OEM ID and OEM Table ID fields in headers of ACPI tables. While it doesn't have impact on default values since QEMU uses 6 and 8 characters long values respectively, it broke usecase where IDs are provided on QEMU CLI. It shouldn't affect guest (but may cause licensing verification issues in guest OS).
One of the broken usecases is user supplied SLIC table with IDs shorter than max possible length, where [2] mangles IDs with extra spaces in RSDT and FADT tables whereas guest OS expects those to mirror the respective values of the used SLIC table.

Fix it by replacing whitespace padding with '\0' padding in accordance with [1] and expectations of guest OS

1) ACPI spec, v2.0b

       17.2 AML Grammar Definition
       ...
       //OEM ID of up to 6 characters. If the OEM ID is
       //shorter than 6 characters, it can be terminated
       //with a NULL character.
Quoted 6 lines

2)

Fixes 602b458201 ("acpi: Permit OEM ID and OEM table ID fields to be changed")

Resolves: https://gitlab.com/qemu-project/qemu/-/issues/707

Quoted 24 lines
---
 hw/acpi/aml-build.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
hw/acpi/aml-build.c +2 -3
index b3b3310df3..65148d5b9d 100644
@@ -1724,9 +1724,9 @@ void acpi_table_begin(AcpiTable *desc, GArray *array)
     build_append_int_noprefix(array, 0, 4); /* Length */
     build_append_int_noprefix(array, desc->rev, 1); /* Revision */
     build_append_int_noprefix(array, 0, 1); /* Checksum */
-    build_append_padded_str(array, desc->oem_id, 6, ' '); /* OEMID */
+    build_append_padded_str(array, desc->oem_id, 6, '\0'); /* OEMID */
     /* OEM Table ID */
-    build_append_padded_str(array, desc->oem_table_id, 8, ' ');
+    build_append_padded_str(array, desc->oem_table_id, 8, '\0');
     build_append_int_noprefix(array, 1, 4); /* OEM Revision */
     g_array_append_vals(array, ACPI_BUILD_APPNAME8, 4); /* Creator ID */
     build_append_int_noprefix(array, 1, 4); /* Creator Revision */
--

2.31.1

#11 lore raw

Re: [PATCH 3/4] acpi: fix OEM ID/OEM Table ID padding

In reply to Igor Mammedov
Quoted 37 lines · Igor Mammedov

Commit [2] broke original '\0' padding of OEM ID and OEM Table ID fields in headers of ACPI tables. While it doesn't have impact on default values since QEMU uses 6 and 8 characters long values respectively, it broke usecase where IDs are provided on QEMU CLI. It shouldn't affect guest (but may cause licensing verification issues in guest OS).
One of the broken usecases is user supplied SLIC table with IDs shorter than max possible length, where [2] mangles IDs with extra spaces in RSDT and FADT tables whereas guest OS expects those to mirror the respective values of the used SLIC table.

Fix it by replacing whitespace padding with '\0' padding in accordance with [1] and expectations of guest OS

1) ACPI spec, v2.0b

        17.2 AML Grammar Definition
        ...
        //OEM ID of up to 6 characters. If the OEM ID is
        //shorter than 6 characters, it can be terminated
        //with a NULL character.

2)

Fixes 602b458201 ("acpi: Permit OEM ID and OEM table ID fields to be changed")

Resolves:https://gitlab.com/qemu-project/qemu/-/issues/707

---
  hw/acpi/aml-build.c | 4 ++--
  1 file changed, 2 insertions(+), 2 deletions(-)
hw/acpi/aml-build.c +0 -0
index b3b3310df3..65148d5b9d 100644
@@ -1724,9 +1724,9 @@ void acpi_table_begin(AcpiTable *desc, GArray *array)
      build_append_int_noprefix(array, 0, 4); /* Length */

hw/acpi/aml-build.c void acpi_table_begin(AcpiTable *desc, GArray *array) · line 1732

Quoted 10 lines
hw/acpi/aml-build.c +2 -2
      build_append_int_noprefix(array, desc->rev, 1); /* Revision */
      build_append_int_noprefix(array, 0, 1); /* Checksum */
-    build_append_padded_str(array, desc->oem_id, 6, ' '); /* OEMID */
+    build_append_padded_str(array, desc->oem_id, 6, '\0'); /* OEMID */
      /* OEM Table ID */
-    build_append_padded_str(array, desc->oem_table_id, 8, ' ');
+    build_append_padded_str(array, desc->oem_table_id, 8, '\0');
      build_append_int_noprefix(array, 1, 4); /* OEM Revision */
      g_array_append_vals(array, ACPI_BUILD_APPNAME8, 4); /* Creator ID */
      build_append_int_noprefix(array, 1, 4); /* Creator Revision */

I can't apply the patch to the qemu-6.1.0 source code on my own.
There is no acpi_table_begin function in the qemu-6.1.0 source code (hw/acpi/aml-buld.c).

#12 lore raw

Re: [PATCH 3/4] acpi: fix OEM ID/OEM Table ID padding

Quoted 4 lines · Dmitry V. Orekhov

I can't apply the patch to the qemu-6.1.0 source code on my own. There is no acpi_table_begin function in the qemu-6.1.0 source code (hw/acpi/aml-buld.c).

Try the following patch :

From 10620c384bf05f0a7561c1afd0ec8ad5af9b7c0f Mon Sep 17 00:00:00 2001
From: Ani Sinha <[email protected]>
Date: Thu, 13 Jan 2022 15:48:16 +0530
Subject: [PATCH] acpi: fix OEM ID/OEM Table ID padding for qemu 6.1.1

Replace whitespace padding with '\0' padding in accordance with spec and expectations of guest OS.

---
 hw/acpi/aml-build.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
hw/acpi/aml-build.c +2 -2
index d5103e6..0df053c 100644
@@ -1703,9 +1703,9 @@ build_header(BIOSLinker *linker, GArray *table_data,
     h->length = cpu_to_le32(len);
     h->revision = rev;
 
-    strpadcpy((char *)h->oem_id, sizeof h->oem_id, oem_id, ' ');
+    strpadcpy((char *)h->oem_id, sizeof h->oem_id, oem_id, '\0');
     strpadcpy((char *)h->oem_table_id, sizeof h->oem_table_id,
-              oem_table_id, ' ');
+              oem_table_id, '\0');
 
     h->oem_revision = cpu_to_le32(1);
     memcpy(h->asl_compiler_id, ACPI_BUILD_APPNAME8, 4);
Signature · 1 line
2.6.3
#13 lore raw

Re: [PATCH 3/4] acpi: fix OEM ID/OEM Table ID padding

Quoted 27 lines · Ani Sinha
Quoted 4 lines · Dmitry V. Orekhov

I can't apply the patch to the qemu-6.1.0 source code on my own. There is no acpi_table_begin function in the qemu-6.1.0 source code (hw/acpi/aml-buld.c).

Try the following patch :

From 10620c384bf05f0a7561c1afd0ec8ad5af9b7c0f Mon Sep 17 00:00:00 2001
From: Ani Sinha <[email protected]>
Date: Thu, 13 Jan 2022 15:48:16 +0530
Subject: [PATCH] acpi: fix OEM ID/OEM Table ID padding for qemu 6.1.1

Replace whitespace padding with '\0' padding in accordance with spec and expectations of guest OS.

---
  hw/acpi/aml-build.c | 4 ++--
  1 file changed, 2 insertions(+), 2 deletions(-)
hw/acpi/aml-build.c +0 -0
index d5103e6..0df053c 100644
@@ -1703,9 +1703,9 @@ build_header(BIOSLinker *linker, GArray *table_data,
      h->length = cpu_to_le32(len);

hw/acpi/aml-build.c build_header(BIOSLinker *linker, GArray *table_data, · line 1711

Quoted 10 lines
hw/acpi/aml-build.c +2 -2
      h->revision = rev;
 
-    strpadcpy((char *)h->oem_id, sizeof h->oem_id, oem_id, ' ');
+    strpadcpy((char *)h->oem_id, sizeof h->oem_id, oem_id, '\0');
      strpadcpy((char *)h->oem_table_id, sizeof h->oem_table_id,
-              oem_table_id, ' ');
+              oem_table_id, '\0');
 
      h->oem_revision = cpu_to_le32(1);
      memcpy(h->asl_compiler_id, ACPI_BUILD_APPNAME8, 4);

The problem has been solved. Thanks.

#14 lore raw

Re: [PATCH 3/4] acpi: fix OEM ID/OEM Table ID padding

In reply to Igor Mammedov
Quoted 15 lines · Igor Mammedov

Commit [2] broke original '\0' padding of OEM ID and OEM Table ID fields in headers of ACPI tables. While it doesn't have impact on default values since QEMU uses 6 and 8 characters long values respectively, it broke usecase where IDs are provided on QEMU CLI. It shouldn't affect guest (but may cause licensing verification issues in guest OS).
One of the broken usecases is user supplied SLIC table with IDs shorter than max possible length, where [2] mangles IDs with extra spaces in RSDT and FADT tables whereas guest OS expects those to mirror the respective values of the used SLIC table.

Fix it by replacing whitespace padding with '\0' padding in accordance with [1] and expectations of guest OS

Quoted 6 lines

1) ACPI spec, v2.0b

       17.2 AML Grammar Definition
       ...
       //OEM ID of up to 6 characters. If the OEM ID is
       //shorter than 6 characters, it can be terminated
       //with a NULL character.

On the other hand, from
https://uefi.org/specs/ACPI/6.4/21_ACPI_Data_Tables_and_Table_Def_Language/ACPI_Data_Tables.html
,

"For example, the OEM ID and OEM Table ID in the common ACPI table header (shown above) are fixed at six and eight characters, respectively. They are not necessarily null terminated"

I also checked version 5 and the verbiage is the same. I think not terminating with a null is not incorrect.

Quoted 30 lines

2)

Fixes 602b458201 ("acpi: Permit OEM ID and OEM table ID fields to be changed")

Resolves: https://gitlab.com/qemu-project/qemu/-/issues/707

---
 hw/acpi/aml-build.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
hw/acpi/aml-build.c +2 -3
index b3b3310df3..65148d5b9d 100644
@@ -1724,9 +1724,9 @@ void acpi_table_begin(AcpiTable *desc, GArray *array)
     build_append_int_noprefix(array, 0, 4); /* Length */
     build_append_int_noprefix(array, desc->rev, 1); /* Revision */
     build_append_int_noprefix(array, 0, 1); /* Checksum */
-    build_append_padded_str(array, desc->oem_id, 6, ' '); /* OEMID */
+    build_append_padded_str(array, desc->oem_id, 6, '\0'); /* OEMID */
     /* OEM Table ID */
-    build_append_padded_str(array, desc->oem_table_id, 8, ' ');
+    build_append_padded_str(array, desc->oem_table_id, 8, '\0');
     build_append_int_noprefix(array, 1, 4); /* OEM Revision */
     g_array_append_vals(array, ACPI_BUILD_APPNAME8, 4); /* Creator ID */
     build_append_int_noprefix(array, 1, 4); /* Creator Revision */
--

2.31.1

#15 lore raw

Re: [PATCH 3/4] acpi: fix OEM ID/OEM Table ID padding

Quoted 27 lines · On Mon, 31 Jan 2022 11:47:00 +0530 Ani Sinha <[email protected]> wrote:
Quoted 15 lines · Igor Mammedov

Commit [2] broke original '\0' padding of OEM ID and OEM Table ID fields in headers of ACPI tables. While it doesn't have impact on default values since QEMU uses 6 and 8 characters long values respectively, it broke usecase where IDs are provided on QEMU CLI. It shouldn't affect guest (but may cause licensing verification issues in guest OS).
One of the broken usecases is user supplied SLIC table with IDs shorter than max possible length, where [2] mangles IDs with extra spaces in RSDT and FADT tables whereas guest OS expects those to mirror the respective values of the used SLIC table.

Fix it by replacing whitespace padding with '\0' padding in accordance with [1] and expectations of guest OS

Quoted 6 lines

1) ACPI spec, v2.0b

       17.2 AML Grammar Definition
       ...
       //OEM ID of up to 6 characters. If the OEM ID is
       //shorter than 6 characters, it can be terminated
       //with a NULL character.  

On the other hand, from
https://uefi.org/specs/ACPI/6.4/21_ACPI_Data_Tables_and_Table_Def_Language/ACPI_Data_Tables.html
,

Quoted 6 lines

"For example, the OEM ID and OEM Table ID in the common ACPI table header (shown above) are fixed at six and eight characters, respectively. They are not necessarily null terminated"

I also checked version 5 and the verbiage is the same. I think not terminating with a null is not incorrect.

I have a trouble with too much 'not' within the sentence. So what's the point of this comment and how it's related to this patch?

Quoted 30 lines
Quoted 29 lines

2)

Fixes 602b458201 ("acpi: Permit OEM ID and OEM table ID fields to be changed")

Resolves: https://gitlab.com/qemu-project/qemu/-/issues/707

---
 hw/acpi/aml-build.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
hw/acpi/aml-build.c +2 -3
index b3b3310df3..65148d5b9d 100644
@@ -1724,9 +1724,9 @@ void acpi_table_begin(AcpiTable *desc, GArray *array)
     build_append_int_noprefix(array, 0, 4); /* Length */
     build_append_int_noprefix(array, desc->rev, 1); /* Revision */
     build_append_int_noprefix(array, 0, 1); /* Checksum */
-    build_append_padded_str(array, desc->oem_id, 6, ' '); /* OEMID */
+    build_append_padded_str(array, desc->oem_id, 6, '\0'); /* OEMID */
     /* OEM Table ID */
-    build_append_padded_str(array, desc->oem_table_id, 8, ' ');
+    build_append_padded_str(array, desc->oem_table_id, 8, '\0');
     build_append_int_noprefix(array, 1, 4); /* OEM Revision */
     g_array_append_vals(array, ACPI_BUILD_APPNAME8, 4); /* Creator ID */
     build_append_int_noprefix(array, 1, 4); /* Creator Revision */
--

2.31.1

#16 lore raw

Re: [PATCH 3/4] acpi: fix OEM ID/OEM Table ID padding

Quoted 30 lines · Igor Mammedov
Quoted 27 lines · On Mon, 31 Jan 2022 11:47:00 +0530 Ani Sinha <[email protected]> wrote:
Quoted 15 lines · Igor Mammedov

Commit [2] broke original '\0' padding of OEM ID and OEM Table ID fields in headers of ACPI tables. While it doesn't have impact on default values since QEMU uses 6 and 8 characters long values respectively, it broke usecase where IDs are provided on QEMU CLI. It shouldn't affect guest (but may cause licensing verification issues in guest OS).
One of the broken usecases is user supplied SLIC table with IDs shorter than max possible length, where [2] mangles IDs with extra spaces in RSDT and FADT tables whereas guest OS expects those to mirror the respective values of the used SLIC table.

Fix it by replacing whitespace padding with '\0' padding in accordance with [1] and expectations of guest OS

Quoted 6 lines

1) ACPI spec, v2.0b

       17.2 AML Grammar Definition
       ...
       //OEM ID of up to 6 characters. If the OEM ID is
       //shorter than 6 characters, it can be terminated
       //with a NULL character.

On the other hand, from
https://uefi.org/specs/ACPI/6.4/21_ACPI_Data_Tables_and_Table_Def_Language/ACPI_Data_Tables.html
,

Quoted 8 lines
Quoted 6 lines

"For example, the OEM ID and OEM Table ID in the common ACPI table header (shown above) are fixed at six and eight characters, respectively. They are not necessarily null terminated"

I also checked version 5 and the verbiage is the same. I think not terminating with a null is not incorrect.

I have a trouble with too much 'not' within the sentence.

:-)

Quoted 2 lines

So what's the point of this comment and how it's related to this patch?

My understanding of the spec is that null termination of both those IDs is not mandatory. Guests may get confused or expect the strings to be null termimated but they should really be open to expecting non-null terminated strings as well. What is important is that the number of chars of those two strings are fixed and well defined in the spec and qemu implementation.

In any case, I think we can leave the patch as is for now and see if the change causes trouble with other guests.

#17 lore raw

Re: [PATCH 3/4] acpi: fix OEM ID/OEM Table ID padding

Quoted 53 lines · On Mon, 31 Jan 2022 18:58:57 +0530 (IST) Ani Sinha <[email protected]> wrote:
Quoted 30 lines · Igor Mammedov
Quoted 27 lines · On Mon, 31 Jan 2022 11:47:00 +0530 Ani Sinha <[email protected]> wrote:
Quoted 15 lines · Igor Mammedov

Commit [2] broke original '\0' padding of OEM ID and OEM Table ID fields in headers of ACPI tables. While it doesn't have impact on default values since QEMU uses 6 and 8 characters long values respectively, it broke usecase where IDs are provided on QEMU CLI. It shouldn't affect guest (but may cause licensing verification issues in guest OS).
One of the broken usecases is user supplied SLIC table with IDs shorter than max possible length, where [2] mangles IDs with extra spaces in RSDT and FADT tables whereas guest OS expects those to mirror the respective values of the used SLIC table.

Fix it by replacing whitespace padding with '\0' padding in accordance with [1] and expectations of guest OS

Quoted 6 lines

1) ACPI spec, v2.0b

       17.2 AML Grammar Definition
       ...
       //OEM ID of up to 6 characters. If the OEM ID is
       //shorter than 6 characters, it can be terminated
       //with a NULL character.  

On the other hand, from
https://uefi.org/specs/ACPI/6.4/21_ACPI_Data_Tables_and_Table_Def_Language/ACPI_Data_Tables.html
,

Quoted 8 lines
Quoted 6 lines

"For example, the OEM ID and OEM Table ID in the common ACPI table header (shown above) are fixed at six and eight characters, respectively. They are not necessarily null terminated"

I also checked version 5 and the verbiage is the same. I think not terminating with a null is not incorrect.

I have a trouble with too much 'not' within the sentence.

:-)

Quoted 2 lines

So what's the point of this comment and how it's related to this patch?

My understanding of the spec is that null termination of both those IDs is not mandatory. Guests may get confused or expect the strings to be null termimated but they should really be open to expecting non-null terminated strings as well. What is important is that the number of chars of those two strings are fixed and well defined in the spec and qemu implementation.

Quoted 2 lines

In any case, I think we can leave the patch as is for now and see if the change causes trouble with other guests.

these fields have a fixed length so one doesn't need terminating NULL in case the full length of the field is utilized, otherwise in case of where the value is shorter than max length it has to be null terminated to express a shorter value. That way QEMU worked for years until 602b458201 introduced regression.

#18 lore raw

Re: [PATCH 3/4] acpi: fix OEM ID/OEM Table ID padding

Quoted 60 lines · Igor Mammedov
Quoted 53 lines · On Mon, 31 Jan 2022 18:58:57 +0530 (IST) Ani Sinha <[email protected]> wrote:
Quoted 30 lines · Igor Mammedov
Quoted 27 lines · On Mon, 31 Jan 2022 11:47:00 +0530 Ani Sinha <[email protected]> wrote:
Quoted 15 lines · Igor Mammedov

Commit [2] broke original '\0' padding of OEM ID and OEM Table ID fields in headers of ACPI tables. While it doesn't have impact on default values since QEMU uses 6 and 8 characters long values respectively, it broke usecase where IDs are provided on QEMU CLI. It shouldn't affect guest (but may cause licensing verification issues in guest OS).
One of the broken usecases is user supplied SLIC table with IDs shorter than max possible length, where [2] mangles IDs with extra spaces in RSDT and FADT tables whereas guest OS expects those to mirror the respective values of the used SLIC table.

Fix it by replacing whitespace padding with '\0' padding in accordance with [1] and expectations of guest OS

Quoted 6 lines

1) ACPI spec, v2.0b

       17.2 AML Grammar Definition
       ...
       //OEM ID of up to 6 characters. If the OEM ID is
       //shorter than 6 characters, it can be terminated
       //with a NULL character.

On the other hand, from
https://uefi.org/specs/ACPI/6.4/21_ACPI_Data_Tables_and_Table_Def_Language/ACPI_Data_Tables.html
,

Quoted 8 lines
Quoted 6 lines

"For example, the OEM ID and OEM Table ID in the common ACPI table header (shown above) are fixed at six and eight characters, respectively. They are not necessarily null terminated"

I also checked version 5 and the verbiage is the same. I think not terminating with a null is not incorrect.

I have a trouble with too much 'not' within the sentence.

:-)

Quoted 2 lines

So what's the point of this comment and how it's related to this patch?

My understanding of the spec is that null termination of both those IDs is not mandatory. Guests may get confused or expect the strings to be null termimated but they should really be open to expecting non-null terminated strings as well. What is important is that the number of chars of those two strings are fixed and well defined in the spec and qemu implementation.

Quoted 2 lines

In any case, I think we can leave the patch as is for now and see if the change causes trouble with other guests.

Quoted 6 lines

these fields have a fixed length so one doesn't need terminating NULL in case the full length of the field is utilized, otherwise in case of where the value is shorter than max length it has to be null terminated to express a shorter value. That way QEMU worked for years until 602b458201 introduced regression.

My comment was based on what I interpreted from reading the latest version of the specs. I guess the spec does not explicitly say what the padding
bytes would be in case the length of the IDs are less the max length. I interpreted the wording to mean that whether or not the
length of the string is shorter, one should not expect it to terminate with null.

It would be nice if a future version of the spec made is explicit and clearer.

#19 lore raw

Re: [PATCH 3/4] acpi: fix OEM ID/OEM Table ID padding

Quoted 69 lines · On Mon, 31 Jan 2022 19:51:24 +0530 (IST) Ani Sinha <[email protected]> wrote:
Quoted 60 lines · Igor Mammedov
Quoted 53 lines · On Mon, 31 Jan 2022 18:58:57 +0530 (IST) Ani Sinha <[email protected]> wrote:
Quoted 30 lines · Igor Mammedov
Quoted 27 lines · On Mon, 31 Jan 2022 11:47:00 +0530 Ani Sinha <[email protected]> wrote:
Quoted 15 lines · Igor Mammedov

Commit [2] broke original '\0' padding of OEM ID and OEM Table ID fields in headers of ACPI tables. While it doesn't have impact on default values since QEMU uses 6 and 8 characters long values respectively, it broke usecase where IDs are provided on QEMU CLI. It shouldn't affect guest (but may cause licensing verification issues in guest OS).
One of the broken usecases is user supplied SLIC table with IDs shorter than max possible length, where [2] mangles IDs with extra spaces in RSDT and FADT tables whereas guest OS expects those to mirror the respective values of the used SLIC table.

Fix it by replacing whitespace padding with '\0' padding in accordance with [1] and expectations of guest OS

Quoted 6 lines

1) ACPI spec, v2.0b

       17.2 AML Grammar Definition
       ...
       //OEM ID of up to 6 characters. If the OEM ID is
       //shorter than 6 characters, it can be terminated
       //with a NULL character.  

On the other hand, from
https://uefi.org/specs/ACPI/6.4/21_ACPI_Data_Tables_and_Table_Def_Language/ACPI_Data_Tables.html
,

Quoted 8 lines
Quoted 6 lines

"For example, the OEM ID and OEM Table ID in the common ACPI table header (shown above) are fixed at six and eight characters, respectively. They are not necessarily null terminated"

I also checked version 5 and the verbiage is the same. I think not terminating with a null is not incorrect.

I have a trouble with too much 'not' within the sentence.

:-)

Quoted 2 lines

So what's the point of this comment and how it's related to this patch?

My understanding of the spec is that null termination of both those IDs is not mandatory. Guests may get confused or expect the strings to be null termimated but they should really be open to expecting non-null terminated strings as well. What is important is that the number of chars of those two strings are fixed and well defined in the spec and qemu implementation.

Quoted 2 lines

In any case, I think we can leave the patch as is for now and see if the change causes trouble with other guests.

Quoted 6 lines

these fields have a fixed length so one doesn't need terminating NULL in case the full length of the field is utilized, otherwise in case of where the value is shorter than max length it has to be null terminated to express a shorter value. That way QEMU worked for years until 602b458201 introduced regression.

Quoted 6 lines

My comment was based on what I interpreted from reading the latest version of the specs. I guess the spec does not explicitly say what the padding
bytes would be in case the length of the IDs are less the max length. I interpreted the wording to mean that whether or not the
length of the string is shorter, one should not expect it to terminate with null.

that's what AML grmamar quoted in commit message clarifies for specific field(s), as opposed to your generic string type description

Quoted 2 lines

It would be nice if a future version of the spec made is explicit and clearer.

PS:
you were asking the other day if there is any bugs left in ACPI, (the answer is that I'm not aware of any).
But there are issues with SMBIOS tables that need to be fixed (it's corner cases with large VM configurations), are you interested in trying to fix it?

#20 lore raw

Re: [PATCH 3/4] acpi: fix OEM ID/OEM Table ID padding

Quoted 79 lines · Igor Mammedov
Quoted 69 lines · On Mon, 31 Jan 2022 19:51:24 +0530 (IST) Ani Sinha <[email protected]> wrote:
Quoted 60 lines · Igor Mammedov
Quoted 53 lines · On Mon, 31 Jan 2022 18:58:57 +0530 (IST) Ani Sinha <[email protected]> wrote:
Quoted 30 lines · Igor Mammedov
Quoted 27 lines · On Mon, 31 Jan 2022 11:47:00 +0530 Ani Sinha <[email protected]> wrote:
Quoted 15 lines · Igor Mammedov

Commit [2] broke original '\0' padding of OEM ID and OEM Table ID fields in headers of ACPI tables. While it doesn't have impact on default values since QEMU uses 6 and 8 characters long values respectively, it broke usecase where IDs are provided on QEMU CLI. It shouldn't affect guest (but may cause licensing verification issues in guest OS).
One of the broken usecases is user supplied SLIC table with IDs shorter than max possible length, where [2] mangles IDs with extra spaces in RSDT and FADT tables whereas guest OS expects those to mirror the respective values of the used SLIC table.

Fix it by replacing whitespace padding with '\0' padding in accordance with [1] and expectations of guest OS

Quoted 6 lines

1) ACPI spec, v2.0b

       17.2 AML Grammar Definition
       ...
       //OEM ID of up to 6 characters. If the OEM ID is
       //shorter than 6 characters, it can be terminated
       //with a NULL character.

On the other hand, from
https://uefi.org/specs/ACPI/6.4/21_ACPI_Data_Tables_and_Table_Def_Language/ACPI_Data_Tables.html
,

Quoted 8 lines
Quoted 6 lines

"For example, the OEM ID and OEM Table ID in the common ACPI table header (shown above) are fixed at six and eight characters, respectively. They are not necessarily null terminated"

I also checked version 5 and the verbiage is the same. I think not terminating with a null is not incorrect.

I have a trouble with too much 'not' within the sentence.

:-)

Quoted 2 lines

So what's the point of this comment and how it's related to this patch?

My understanding of the spec is that null termination of both those IDs is not mandatory. Guests may get confused or expect the strings to be null termimated but they should really be open to expecting non-null terminated strings as well. What is important is that the number of chars of those two strings are fixed and well defined in the spec and qemu implementation.

Quoted 2 lines

In any case, I think we can leave the patch as is for now and see if the change causes trouble with other guests.

Quoted 6 lines

these fields have a fixed length so one doesn't need terminating NULL in case the full length of the field is utilized, otherwise in case of where the value is shorter than max length it has to be null terminated to express a shorter value. That way QEMU worked for years until 602b458201 introduced regression.

Quoted 6 lines

My comment was based on what I interpreted from reading the latest version of the specs. I guess the spec does not explicitly say what the padding
bytes would be in case the length of the IDs are less the max length. I interpreted the wording to mean that whether or not the
length of the string is shorter, one should not expect it to terminate with null.

Quoted 3 lines

that's what AML grmamar quoted in commit message clarifies for specific field(s), as opposed to your generic string type description

Ah yes, my bad. In
https://uefi.org/specs/ACPI/6.4/20_AML_Specification/AML_Specification.html , section 20.2.1 has this also :

ByteData(6) // OEM ID of up to 6 characters. If the OEM ID is shorter than 6 characters,
it can be terminated with a NULL character.

etc. Somehow I missed it.

Quoted 4 lines

PS:
you were asking the other day if there is any bugs left in ACPI, (the answer is that I'm not aware of any).

Yes spoke to Gerd offline. On native side also he is unaware of any issues post 6.2.

Quoted 3 lines

But there are issues with SMBIOS tables that need to be fixed (it's corner cases with large VM configurations), are you interested in trying to fix it?

Yes sure. I will try my best.

#21 lore raw

Re: [PATCH 3/4] acpi: fix OEM ID/OEM Table ID padding

Quoted 99 lines · On Tue, 1 Feb 2022 13:25:20 +0530 (IST) Ani Sinha <[email protected]> wrote:
Quoted 79 lines · Igor Mammedov
Quoted 69 lines · On Mon, 31 Jan 2022 19:51:24 +0530 (IST) Ani Sinha <[email protected]> wrote:
Quoted 60 lines · Igor Mammedov
Quoted 53 lines · On Mon, 31 Jan 2022 18:58:57 +0530 (IST) Ani Sinha <[email protected]> wrote:
Quoted 30 lines · Igor Mammedov
Quoted 27 lines · On Mon, 31 Jan 2022 11:47:00 +0530 Ani Sinha <[email protected]> wrote:
Quoted 15 lines · Igor Mammedov

Commit [2] broke original '\0' padding of OEM ID and OEM Table ID fields in headers of ACPI tables. While it doesn't have impact on default values since QEMU uses 6 and 8 characters long values respectively, it broke usecase where IDs are provided on QEMU CLI. It shouldn't affect guest (but may cause licensing verification issues in guest OS).
One of the broken usecases is user supplied SLIC table with IDs shorter than max possible length, where [2] mangles IDs with extra spaces in RSDT and FADT tables whereas guest OS expects those to mirror the respective values of the used SLIC table.

Fix it by replacing whitespace padding with '\0' padding in accordance with [1] and expectations of guest OS

Quoted 6 lines

1) ACPI spec, v2.0b

       17.2 AML Grammar Definition
       ...
       //OEM ID of up to 6 characters. If the OEM ID is
       //shorter than 6 characters, it can be terminated
       //with a NULL character.  

On the other hand, from
https://uefi.org/specs/ACPI/6.4/21_ACPI_Data_Tables_and_Table_Def_Language/ACPI_Data_Tables.html
,

Quoted 8 lines
Quoted 6 lines

"For example, the OEM ID and OEM Table ID in the common ACPI table header (shown above) are fixed at six and eight characters, respectively. They are not necessarily null terminated"

I also checked version 5 and the verbiage is the same. I think not terminating with a null is not incorrect.

I have a trouble with too much 'not' within the sentence.

:-)

Quoted 2 lines

So what's the point of this comment and how it's related to this patch?

My understanding of the spec is that null termination of both those IDs is not mandatory. Guests may get confused or expect the strings to be null termimated but they should really be open to expecting non-null terminated strings as well. What is important is that the number of chars of those two strings are fixed and well defined in the spec and qemu implementation.

Quoted 2 lines

In any case, I think we can leave the patch as is for now and see if the change causes trouble with other guests.

Quoted 6 lines

these fields have a fixed length so one doesn't need terminating NULL in case the full length of the field is utilized, otherwise in case of where the value is shorter than max length it has to be null terminated to express a shorter value. That way QEMU worked for years until 602b458201 introduced regression.

Quoted 6 lines

My comment was based on what I interpreted from reading the latest version of the specs. I guess the spec does not explicitly say what the padding
bytes would be in case the length of the IDs are less the max length. I interpreted the wording to mean that whether or not the
length of the string is shorter, one should not expect it to terminate with null.

Quoted 3 lines

that's what AML grmamar quoted in commit message clarifies for specific field(s), as opposed to your generic string type description

Ah yes, my bad. In
https://uefi.org/specs/ACPI/6.4/20_AML_Specification/AML_Specification.html , section 20.2.1 has this also :

ByteData(6) // OEM ID of up to 6 characters. If the OEM ID is shorter than 6 characters,
it can be terminated with a NULL character.

etc. Somehow I missed it.

Quoted 4 lines

PS:
you were asking the other day if there is any bugs left in ACPI, (the answer is that I'm not aware of any).

Quoted 9 lines

Yes spoke to Gerd offline. On native side also he is unaware of any issues post 6.2.

Quoted 3 lines

But there are issues with SMBIOS tables that need to be fixed (it's corner cases with large VM configurations), are you interested in trying to fix it?

Yes sure. I will try my best.

Thanks,
here is SMBIOS corruption bug from backlog, created by Eduardo when he was investigating the issue
https://bugzilla.redhat.com/show_bug.cgi?id=2023977

#22 lore raw

PATCH 4/4 tests: acpi: update expected blobs

In reply to Igor Mammedov

Expected changes caused by previous commit:

nvdimm ssdt (q35/pc/virt):

  - *     OEM Table ID     "NVDIMM  "
  + *     OEM Table ID     "NVDIMM"

SLIC test FADT (tests/data/acpi/q35/FACP.slic):

  -[010h 0016   8]                 Oem Table ID : "ME      "
  +[010h 0016   8]                 Oem Table ID : "ME"
---
 tests/qtest/bios-tables-test-allowed-diff.h |   4 ----
 tests/data/acpi/pc/SSDT.dimmpxm             | Bin 734 -> 734 bytes
 tests/data/acpi/q35/FACP.slic               | Bin 244 -> 244 bytes
 tests/data/acpi/q35/SSDT.dimmpxm            | Bin 734 -> 734 bytes
 tests/data/acpi/virt/SSDT.memhp             | Bin 736 -> 736 bytes
 5 files changed, 4 deletions(-)
tests/qtest/bios-tables-test-allowed-diff.h +0 -4
index 7faa8f53be..dfb8523c8b 100644
@@ -1,5 +1 @@
 /* List of comma-separated changed AML files to ignore */
-"tests/data/acpi/virt/SSDT.memhp",
-"tests/data/acpi/pc/SSDT.dimmpxm",
-"tests/data/acpi/q35/SSDT.dimmpxm",
-"tests/data/acpi/q35/FACP.slic",
tests/data/acpi/pc/SSDT.dimmpxm binary +0 -0
index a50a961fa1d9b0dd8ea4096d652c83bcf04db20b..ac55387d57e48adb99eb738a102308688a262fb8 100644
GIT binary patch
tests/data/acpi/q35/FACP.slic binary +0 -0
index 891fd4b784b7b6b3ea303976db7ecd5b669bc84b..15986e095cf2db7ee92f7ce113c1d46d54018c62 100644
GIT binary patch
tests/data/acpi/q35/SSDT.dimmpxm binary +0 -0
index 617a1c911c7d6753bcedc8ecc52e3027a5259ad6..98e6f0e3f3bb02dd419e36bdd1db9b94c728c406 100644
GIT binary patch
tests/data/acpi/virt/SSDT.memhp binary +0 -0
index e8b850ae2239d8f496b12de672c2a1268e2f269d..375d7b6fc85a484f492a26ccd355c205f2c34473 100644
GIT binary patch
Signature · 1 line
2.31.1
Acked-by on v1 4/4
#23 lore raw

Re: [PATCH 4/4] tests: acpi: update expected blobs

Quoted 11 lines · Igor Mammedov

Expected changes caused by previous commit:

nvdimm ssdt (q35/pc/virt):

  - *     OEM Table ID     "NVDIMM  "
  + *     OEM Table ID     "NVDIMM"

SLIC test FADT (tests/data/acpi/q35/FACP.slic):

  -[010h 0016   8]                 Oem Table ID : "ME      "
  +[010h 0016   8]                 Oem Table ID : "ME"
Quoted 58 lines
---
 tests/qtest/bios-tables-test-allowed-diff.h |   4 ----
 tests/data/acpi/pc/SSDT.dimmpxm             | Bin 734 -> 734 bytes
 tests/data/acpi/q35/FACP.slic               | Bin 244 -> 244 bytes
 tests/data/acpi/q35/SSDT.dimmpxm            | Bin 734 -> 734 bytes
 tests/data/acpi/virt/SSDT.memhp             | Bin 736 -> 736 bytes
 5 files changed, 4 deletions(-)
tests/qtest/bios-tables-test-allowed-diff.h +0 -4
index 7faa8f53be..dfb8523c8b 100644
@@ -1,5 +1 @@
 /* List of comma-separated changed AML files to ignore */
-"tests/data/acpi/virt/SSDT.memhp",
-"tests/data/acpi/pc/SSDT.dimmpxm",
-"tests/data/acpi/q35/SSDT.dimmpxm",
-"tests/data/acpi/q35/FACP.slic",
tests/data/acpi/pc/SSDT.dimmpxm binary +0 -0
index a50a961fa1d9b0dd8ea4096d652c83bcf04db20b..ac55387d57e48adb99eb738a102308688a262fb8 100644
GIT binary patch
tests/data/acpi/q35/FACP.slic binary +0 -0
index 891fd4b784b7b6b3ea303976db7ecd5b669bc84b..15986e095cf2db7ee92f7ce113c1d46d54018c62 100644
GIT binary patch
tests/data/acpi/q35/SSDT.dimmpxm binary +0 -0
index 617a1c911c7d6753bcedc8ecc52e3027a5259ad6..98e6f0e3f3bb02dd419e36bdd1db9b94c728c406 100644
GIT binary patch
tests/data/acpi/virt/SSDT.memhp binary +0 -1
index e8b850ae2239d8f496b12de672c2a1268e2f269d..375d7b6fc85a484f492a26ccd355c205f2c34473 100644
GIT binary patch
 
--

2.31.1

#24 lore raw

[PATCH 5/4] tests: acpi: test short OEM_ID/OEM_TABLE_ID values in test_oem_fields()

In reply to Igor Mammedov

Previous patch [1] added explicit whitespace padding to OEM_ID/OEM_TABLE_ID values used in test_oem_fields() testcase to avoid false positive and bisection issues when QEMU is switched to \0' padding. As result testcase ceased to test values that were shorter than max possible length values.

Update testcase to make sure that it's testing shorter IDs like it used to before [2].

1) "tests: acpi: manually pad OEM_ID/OEM_TABLE_ID for test_oem_fields() test"
2) 602b458201 ("acpi: Permit OEM ID and OEM table ID fields to be changed")

---
 tests/qtest/bios-tables-test.c | 12 ++++++------
 1 file changed, 6 insertions(+), 6 deletions(-)
tests/qtest/bios-tables-test.c +6 -6
index 90c9f6a0a2..ad536fd7b1 100644
@@ -71,10 +71,10 @@
 
 #define ACPI_REBUILD_EXPECTED_AML "TEST_ACPI_REBUILD_AML"
 
-#define OEM_ID             "TEST  "
-#define OEM_TABLE_ID       "OEM     "
-#define OEM_TEST_ARGS      "-machine x-oem-id='" OEM_ID "',x-oem-table-id='" \
-                           OEM_TABLE_ID "'"
+#define OEM_ID             "TEST"
+#define OEM_TABLE_ID       "OEM"
+#define OEM_TEST_ARGS      "-machine x-oem-id=" OEM_ID ",x-oem-table-id=" \
+                           OEM_TABLE_ID
 
 typedef struct {
     bool tcg_only;
@@ -1530,8 +1530,8 @@ static void test_oem_fields(test_data *data)
             continue;
         }
 
-        g_assert(memcmp(sdt->aml + 10, OEM_ID, 6) == 0);
-        g_assert(memcmp(sdt->aml + 16, OEM_TABLE_ID, 8) == 0);
+        g_assert(strncmp((char *)sdt->aml + 10, OEM_ID, 6) == 0);
+        g_assert(strncmp((char *)sdt->aml + 16, OEM_TABLE_ID, 8) == 0);
     }
 }
 
Signature · 1 line
2.31.1
Reviewed-by
#25 lore raw

Re: [PATCH 5/4] tests: acpi: test short OEM_ID/OEM_TABLE_ID values in test_oem_fields()

Quoted 6 lines · Igor Mammedov

Previous patch [1] added explicit whitespace padding to OEM_ID/OEM_TABLE_ID values used in test_oem_fields() testcase to avoid false positive and bisection issues when QEMU is switched to \0' padding. As result testcase ceased to test values that were shorter than max possible length values.

Quoted 8 lines

Update testcase to make sure that it's testing shorter IDs like it used to before [2].

1) "tests: acpi: manually pad OEM_ID/OEM_TABLE_ID for test_oem_fields() test"
2) 602b458201 ("acpi: Permit OEM ID and OEM table ID fields to be changed")

Quoted 1 line
#26 lore raw
In reply to Igor Mammedov
Quoted 8 lines · On Wed, 12 Jan 2022 08:03:28 -0500 Igor Mammedov <[email protected]> wrote:

Since 6.0 the commit:
602b458201 ("acpi: Permit OEM ID and OEM table ID fields to be changed")
regressed values of OEM [Table] ID fields in ACPI tables
by padding them with whitespace is a value is shorter then
max possible. That depending on vendor broke OEM [Table] ID patching with SLIC table values and as result licensing of Windows guests.

First reported here https://gitlab.com/qemu-project/qemu/-/issues/707

ping,
Michael can you pick it up so that downstreams could backport the fix?

Quoted 20 lines
CC Marian Postevca <[email protected]>
CC Michael S. Tsirkin <[email protected]>
CC Ani Sinha <[email protected]>

Igor Mammedov (4):
tests: acpi: manually pad OEM_ID/OEM_TABLE_ID for test_oem_fields()
test
tests: acpi: whitelist nvdimm's SSDT and FACP.slic expected blobs
acpi: fix OEM ID/OEM Table ID padding
tests: acpi: update expected blobs

 hw/acpi/aml-build.c              |   4 ++--
 tests/data/acpi/pc/SSDT.dimmpxm  | Bin 734 -> 734 bytes
 tests/data/acpi/q35/FACP.slic    | Bin 244 -> 244 bytes
 tests/data/acpi/q35/SSDT.dimmpxm | Bin 734 -> 734 bytes
 tests/data/acpi/virt/SSDT.memhp  | Bin 736 -> 736 bytes
 tests/qtest/bios-tables-test.c   |  15 ++++++---------
 6 files changed, 8 insertions(+), 11 deletions(-)
#27 lore raw
Quoted 10 lines · Igor Mammedov
Quoted 7 lines · On Wed, 12 Jan 2022 08:03:28 -0500 Igor Mammedov <[email protected]> wrote:

Since 6.0 the commit:
602b458201 ("acpi: Permit OEM ID and OEM table ID fields to be changed")
regressed values of OEM [Table] ID fields in ACPI tables
by padding them with whitespace is a value is shorter then
max possible. That depending on vendor broke OEM [Table] ID patching with SLIC table values and as result licensing of Windows guests.

Quoted 5 lines
Quoted 1 line

First reported here https://gitlab.com/qemu-project/qemu/-/issues/707

ping,
Michael can you pick it up so that downstreams could backport the fix?

I see the differences with Ani have been resolved, will be in the next pull.

Quoted 20 lines
Quoted 20 lines
CC Marian Postevca <[email protected]>
CC Michael S. Tsirkin <[email protected]>
CC Ani Sinha <[email protected]>

Igor Mammedov (4):
tests: acpi: manually pad OEM_ID/OEM_TABLE_ID for test_oem_fields()
test
tests: acpi: whitelist nvdimm's SSDT and FACP.slic expected blobs
acpi: fix OEM ID/OEM Table ID padding
tests: acpi: update expected blobs

 hw/acpi/aml-build.c              |   4 ++--
 tests/data/acpi/pc/SSDT.dimmpxm  | Bin 734 -> 734 bytes
 tests/data/acpi/q35/FACP.slic    | Bin 244 -> 244 bytes
 tests/data/acpi/q35/SSDT.dimmpxm | Bin 734 -> 734 bytes
 tests/data/acpi/virt/SSDT.memhp  | Bin 736 -> 736 bytes
 tests/qtest/bios-tables-test.c   |  15 ++++++---------
 6 files changed, 8 insertions(+), 11 deletions(-)