untrusted comment: verify with openbsd-73-base.pub
RWQS90bYzZ4XFsOCTOc6JNPHLc38KC/XGFOkWX2rTn+wRs2+io7y6ZcEWDR7nX3Wmp75hnHYCCEwbmGemK6hVd8z0GtRS1OX+Qg=

OpenBSD 7.3 errata 013, July 24, 2023:

Install firmware updates for AMD cpus, to repair past or potential future bugs.
For i386 and amd64 platforms.

Apply by doing:
   signify -Vep /etc/signify/openbsd-73-base.pub -x 013_amdcpufirmware.patch.sig \
       -m - | (cd /usr/src && patch -p0)

And then rebuild and install a new kernel:
   KK=`sysctl -n kern.osversion | cut -d# -f1`
   cd /usr/src/sys/arch/`machine`/compile/$KK
   make obj
   make config
   make
   make install

New firmwares can be downloaded using
   fw_update

New bootblocks will need to be installed, using a command similar to this:
   installboot -v sd0

Index: sys/arch/amd64/amd64/ucode.c
===================================================================
RCS file: /cvs/src/sys/arch/amd64/amd64/ucode.c,v
diff -u -p -u -r1.5 ucode.c
--- sys/arch/amd64/amd64/ucode.c        28 Jun 2019 21:54:05 -0000      1.5
+++ sys/arch/amd64/amd64/ucode.c        23 Jul 2023 02:36:41 -0000
@@ -72,6 +72,8 @@ size_t         cpu_ucode_size;
void    cpu_ucode_setup(void);
void    cpu_ucode_apply(struct cpu_info *);

+struct mutex   cpu_ucode_mtx = MUTEX_INITIALIZER(IPL_HIGH);
+
/* Intel */
void    cpu_ucode_intel_apply(struct cpu_info *);
struct intel_ucode_header *
@@ -82,7 +84,8 @@ int    cpu_ucode_intel_match(struct intel_
uint32_t cpu_ucode_intel_rev(void);

struct intel_ucode_header      *cpu_ucode_intel_applied;
-struct mutex                    cpu_ucode_intel_mtx = MUTEX_INITIALIZER(IPL_HIGH);
+
+void cpu_ucode_amd_apply(struct cpu_info *);

void
cpu_ucode_setup(void)
@@ -107,6 +110,99 @@ cpu_ucode_apply(struct cpu_info *ci)
{
       if (strcmp(cpu_vendor, "GenuineIntel") == 0)
               cpu_ucode_intel_apply(ci);
+       else if (strcmp(cpu_vendor, "AuthenticAMD") == 0)
+               cpu_ucode_amd_apply(ci);
+}
+
+#define AMD_MAGIC 0x00414d44
+
+struct amd_equiv {
+       uint32_t id;
+       uint32_t a;
+       uint32_t b;
+       uint16_t eid;
+       uint16_t c;
+} __packed;
+
+struct amd_patch {
+       uint32_t type;
+       uint32_t len;
+       uint32_t a;
+       uint32_t level;
+       uint8_t c[16];
+       uint16_t eid;
+} __packed;
+
+void
+cpu_ucode_amd_apply(struct cpu_info *ci)
+{
+       uint64_t level;
+       uint32_t magic, tlen, i;
+       uint16_t eid = 0;
+       uint32_t sig, ebx, ecx, edx;
+       uint64_t start = 0;
+
+       if (cpu_ucode_data == NULL || cpu_ucode_size == 0) {
+               DPRINTF(("%s: no microcode provided\n", __func__));
+               return;
+       }
+
+       /*
+        * Grab a mutex, because we are not allowed to run updates
+        * simultaneously on HT siblings.
+        */
+       mtx_enter(&cpu_ucode_mtx);
+
+       CPUID(1, sig, ebx, ecx, edx);
+
+       level = rdmsr(MSR_PATCH_LEVEL);
+       DPRINTF(("%s: cur patch level 0x%llx\n", __func__, level));
+
+       memcpy(&magic, cpu_ucode_data, 4);
+       if (magic != AMD_MAGIC) {
+               DPRINTF(("%s: bad magic %x\n", __func__, magic));
+               goto out;
+       }
+
+       memcpy(&tlen, &cpu_ucode_data[8], 4);
+
+       /* find equivalence id matching our cpu signature */
+       for (i = 12; i < 12 + tlen;) {
+               struct amd_equiv ae;
+               if (i + sizeof(ae) > cpu_ucode_size) {
+                       DPRINTF(("%s: truncated etable\n", __func__));
+                       goto out;
+               }
+               memcpy(&ae, &cpu_ucode_data[i], sizeof(ae));
+               i += sizeof(ae);
+               if (ae.id == sig)
+                       eid = ae.eid;
+       }
+
+       /* look for newer patch with the equivalence id */
+       while (i < cpu_ucode_size) {
+               struct amd_patch ap;
+               if (i + sizeof(ap) > cpu_ucode_size) {
+                       DPRINTF(("%s: truncated ptable\n", __func__));
+                       goto out;
+               }
+               memcpy(&ap, &cpu_ucode_data[i], sizeof(ap));
+               if (ap.type == 1 && ap.eid == eid && ap.level > level)
+                       start = (uint64_t)&cpu_ucode_data[i + 8];
+               if (i + ap.len + 8 > cpu_ucode_size) {
+                       DPRINTF(("%s: truncated patch\n", __func__));
+                       goto out;
+               }
+               i += ap.len + 8;
+       }
+
+       if (start != 0) {
+               wrmsr(MSR_PATCH_LOADER, start);
+               level = rdmsr(MSR_PATCH_LEVEL);
+               DPRINTF(("%s: new patch level 0x%llx\n", __func__, level));
+       }
+out:
+       mtx_leave(&cpu_ucode_mtx);
}

void
@@ -125,7 +221,7 @@ cpu_ucode_intel_apply(struct cpu_info *c
        * Grab a mutex, because we are not allowed to run updates
        * simultaneously on HT siblings.
        */
-       mtx_enter(&cpu_ucode_intel_mtx);
+       mtx_enter(&cpu_ucode_mtx);

       old_rev = cpu_ucode_intel_rev();
       update = cpu_ucode_intel_applied;
@@ -159,7 +255,7 @@ cpu_ucode_intel_apply(struct cpu_info *c
       }

out:
-       mtx_leave(&cpu_ucode_intel_mtx);
+       mtx_leave(&cpu_ucode_mtx);
}

struct intel_ucode_header *
Index: sys/arch/i386/i386/ucode.c
===================================================================
RCS file: /cvs/src/sys/arch/i386/i386/ucode.c,v
diff -u -p -u -r1.3 ucode.c
--- sys/arch/i386/i386/ucode.c  15 Dec 2020 22:51:34 -0000      1.3
+++ sys/arch/i386/i386/ucode.c  23 Jul 2023 02:53:41 -0000
@@ -72,6 +72,8 @@ size_t         cpu_ucode_size;
void    cpu_ucode_setup(void);
void    cpu_ucode_apply(struct cpu_info *);

+struct mutex   cpu_ucode_mtx = MUTEX_INITIALIZER(IPL_HIGH);
+
/* Intel */
void    cpu_ucode_intel_apply(struct cpu_info *);
struct intel_ucode_header *
@@ -82,7 +84,8 @@ int    cpu_ucode_intel_match(struct intel_
uint32_t cpu_ucode_intel_rev(void);

struct intel_ucode_header      *cpu_ucode_intel_applied;
-struct mutex                    cpu_ucode_intel_mtx = MUTEX_INITIALIZER(IPL_HIGH);
+
+void cpu_ucode_amd_apply(struct cpu_info *);

void
cpu_ucode_setup(void)
@@ -130,6 +133,99 @@ cpu_ucode_apply(struct cpu_info *ci)
{
       if (strcmp(cpu_vendor, "GenuineIntel") == 0)
               cpu_ucode_intel_apply(ci);
+       else if (strcmp(cpu_vendor, "AuthenticAMD") == 0)
+               cpu_ucode_amd_apply(ci);
+}
+
+#define AMD_MAGIC 0x00414d44
+
+struct amd_equiv {
+       uint32_t id;
+       uint32_t a;
+       uint32_t b;
+       uint16_t eid;
+       uint16_t c;
+} __packed;
+
+struct amd_patch {
+       uint32_t type;
+       uint32_t len;
+       uint32_t a;
+       uint32_t level;
+       uint8_t c[16];
+       uint16_t eid;
+} __packed;
+
+void
+cpu_ucode_amd_apply(struct cpu_info *ci)
+{
+       uint64_t level;
+       uint32_t magic, tlen, i;
+       uint16_t eid = 0;
+       uint32_t sig, ebx, ecx, edx;
+       uint64_t start = 0;
+
+       if (cpu_ucode_data == NULL || cpu_ucode_size == 0) {
+               DPRINTF(("%s: no microcode provided\n", __func__));
+               return;
+       }
+
+       /*
+        * Grab a mutex, because we are not allowed to run updates
+        * simultaneously on HT siblings.
+        */
+       mtx_enter(&cpu_ucode_mtx);
+
+       CPUID(1, sig, ebx, ecx, edx);
+
+       level = rdmsr(MSR_PATCH_LEVEL);
+       DPRINTF(("%s: cur patch level 0x%llx\n", __func__, level));
+
+       memcpy(&magic, cpu_ucode_data, 4);
+       if (magic != AMD_MAGIC) {
+               DPRINTF(("%s: bad magic %x\n", __func__, magic));
+               goto out;
+       }
+
+       memcpy(&tlen, &cpu_ucode_data[8], 4);
+
+       /* find equivalence id matching our cpu signature */
+       for (i = 12; i < 12 + tlen;) {
+               struct amd_equiv ae;
+               if (i + sizeof(ae) > cpu_ucode_size) {
+                       DPRINTF(("%s: truncated etable\n", __func__));
+                       goto out;
+               }
+               memcpy(&ae, &cpu_ucode_data[i], sizeof(ae));
+               i += sizeof(ae);
+               if (ae.id == sig)
+                       eid = ae.eid;
+       }
+
+       /* look for newer patch with the equivalence id */
+       while (i < cpu_ucode_size) {
+               struct amd_patch ap;
+               if (i + sizeof(ap) > cpu_ucode_size) {
+                       DPRINTF(("%s: truncated ptable\n", __func__));
+                       goto out;
+               }
+               memcpy(&ap, &cpu_ucode_data[i], sizeof(ap));
+               if (ap.type == 1 && ap.eid == eid && ap.level > level)
+                       start = (uint64_t)&cpu_ucode_data[i + 8];
+               if (i + ap.len + 8 > cpu_ucode_size) {
+                       DPRINTF(("%s: truncated patch\n", __func__));
+                       goto out;
+               }
+               i += ap.len + 8;
+       }
+
+       if (start != 0) {
+               wrmsr(MSR_PATCH_LOADER, start);
+               level = rdmsr(MSR_PATCH_LEVEL);
+               DPRINTF(("%s: new patch level 0x%llx\n", __func__, level));
+       }
+out:
+       mtx_leave(&cpu_ucode_mtx);
}

void
@@ -148,7 +244,7 @@ cpu_ucode_intel_apply(struct cpu_info *c
        * Grab a mutex, because we are not allowed to run updates
        * simultaneously on HT siblings.
        */
-       mtx_enter(&cpu_ucode_intel_mtx);
+       mtx_enter(&cpu_ucode_mtx);

       old_rev = cpu_ucode_intel_rev();
       update = cpu_ucode_intel_applied;
@@ -182,7 +278,7 @@ cpu_ucode_intel_apply(struct cpu_info *c
       }

out:
-       mtx_leave(&cpu_ucode_intel_mtx);
+       mtx_leave(&cpu_ucode_mtx);
}

struct intel_ucode_header *

You are viewing proxied material from ftp.icm.edu.pl. The copyright of proxied material belongs to its original authors. Any comments or complaints in relation to proxied material should be directed to the original authors of the content concerned. Please see the disclaimer for more details.