coreboot-kgpe-d16/util/inteltool/pcie.c

474 lines
16 KiB
C
Raw Normal View History

/*
* inteltool - dump all registers on an Intel CPU + chipset based system.
*
* Copyright (C) 2008-2010 by coresystems GmbH
* Copyright (C) 2012 Anton Kochkov
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; version 2 of the License.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, write to the Free Software
* Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
*/
#include <stdio.h>
#include <stdlib.h>
#include <inttypes.h>
#include "inteltool.h"
/* 320766 */
static const io_register_t nehalem_dmi_registers[] = {
{ 0x00, 4, "DMIVCH" }, // DMI Virtual Channel Capability Header
{ 0x04, 4, "DMIVCCAP1" }, // DMI Port VC Capability Register 1
{ 0x08, 4, "DMIVCCAP2" }, // DMI Port VC Capability Register 2
{ 0x0C, 4, "DMIVCCTL" }, // DMI Port VC Control
{ 0x10, 4, "DMIVC0RCAP" }, // DMI VC0 Resource Capability
{ 0x14, 4, "DMIVC0RCTL" }, // DMI VC0 Resource Control
/* { 0x18, 2, "RSVD" }, // Reserved */
{ 0x1A, 2, "DMIVC0RSTS" }, // DMI VC0 Resource Status
{ 0x1C, 4, "DMIVC1RCAP" }, // DMI VC1 Resource Capability
{ 0x20, 4, "DMIVC1RCTL" }, // DMI VC1 Resource Control
/* { 0x24, 2, "RSVD" }, // Reserved */
{ 0x26, 2, "DMIVC1RSTS" }, // DMI VC1 Resource Status
/* ... - Reserved */
{ 0x84, 4, "DMILCAP" }, // DMI Link Capabilities
{ 0x88, 2, "DMILCTL" }, // DMI Link Control
{ 0x8A, 2, "DMILSTS" }, // DMI Link Status
/* ... - Reserved */
};
/* 322812 */
static const io_register_t westmere_dmi_registers[] = {
{ 0x00, 4, "DMIVCECH" }, // DMI Virtual Channel Enhanced Capability
{ 0x04, 4, "DMIPVCCAP1" }, // DMI Port VC Capability Register 1
{ 0x08, 4, "DMIPVCCAP2" }, // DMI Port VC Capability Register 2
{ 0x0C, 2, "DMIPVCCTL" }, // DMI Port VC Control
/* { 0x0E, 2, "RSVD" }, // Reserved */
{ 0x10, 4, "DMIVC0RCAP" }, // DMI VC0 Resource Capability
{ 0x14, 4, "DMIVC0RCTL" }, // DMI VC0 Resource Control
/* { 0x18, 2, "RSVD" }, // Reserved */
{ 0x1A, 2, "DMIVC0RSTS" }, // DMI VC0 Resource Status
{ 0x1C, 4, "DMIVC1RCAP" }, // DMI VC1 Resource Capability
{ 0x20, 4, "DMIVC1RCTL1" }, // DMI VC1 Resource Control
/* { 0x24, 2, "RSVD" }, // Reserved */
{ 0x26, 2, "DMIC1RSTS" }, // DMI VC1 Resource Status
/* ... - Reserved */
{ 0x84, 4, "DMILCAP" }, // DMI Link Capabilities
{ 0x88, 2, "DMILCTL" }, // DMI Link Control
{ 0x8A, 2, "DMILSTS" }, // DMI Link Status
/* ... - Reserved */
};
static const io_register_t sandybridge_dmi_registers[] = {
{ 0x00, 4, "DMI VCECH" }, // DMI Virtual Channel Enhanced Capability
{ 0x04, 4, "DMI PVCCAP1" }, // DMI Port VC Capability Register 1
{ 0x08, 4, "DMI PVVAP2" }, // DMI Port VC Capability Register 2
{ 0x0C, 2, "DMI PVCCTL" }, // DMI Port VC Control
/* { 0x0E, 2, "RSVD" }, // Reserved */
{ 0x10, 4, "DMI VC0RCAP" }, // DMI VC0 Resource Capability
{ 0x14, 4, "DMI VC0RCTL" }, // DMI VC0 Resource Control
/* { 0x18, 2, "RSVD" }, // Reserved */
{ 0x1A, 2, "DMI VC0RSTS" }, // DMI VC0 Resource Status
{ 0x1C, 4, "DMI VC1RCAP" }, // DMI VC1 Resource Capability
{ 0x20, 4, "DMI VC1RCTL" }, // DMI VC1 Resource Control
/* { 0x24, 2, "RSVD" }, // Reserved */
{ 0x26, 2, "DMI VC1RSTS" }, // DMI VC1 Resource Status
{ 0x28, 4, "DMI VCPRCAP" }, // DMI VCp Resource Capability
{ 0x2C, 4, "DMI VCPRCTL" }, // DMI VCp Resource Control
/* { 0x30, 2, "RSVD" }, // Reserved */
{ 0x32, 2, "DMI VCPRSTS" }, // DMI VCp Resource Status
{ 0x34, 4, "DMI VCMRCAP" }, // DMI VCm Resource Capability
{ 0x38, 4, "DMI VCMRCTL" }, // DMI VCm Resource Control
/* { 0x3C, 2, "RSVD" }, // Reserved */
{ 0x3E, 2, "DMI VCMRSTS" }, // DMI VCm Resource Status
/* { 0x40, 4, "RSVD" }, // Reserved */
{ 0x44, 4, "DMI ESC" }, // DMI Element Self Description
/* { 0x48, 8, "RSVD" }, // Reserved */
{ 0x50, 4, "DMI LE1D" }, // DMI Link Entry 1 Description
/* { 0x54, 4, "RSVD" }, // Reserved */
{ 0x58, 4, "DMI LE1A" }, // DMI Link Entry 1 Address
{ 0x5C, 4, "DMI LUE1A" }, // DMI Link Upper Entry 1 Address
{ 0x60, 4, "DMI LE2D" }, // DMI Link Entry 2 Description
/* { 0x64, 4, "RSVD" }, // Reserved */
{ 0x68, 4, "DMI LE2A" }, // DMI Link Entry 2 Address
/* { 0x6C, 4, "RSVD" }, // Reserved
{ 0x70, 8, "RSVD" }, // Reserved
{ 0x78, 8, "RSVD" }, // Reserved
{ 0x80, 4, "RSVD" }, // Reserved */
{ 0x84, 4, "LCAP" }, // Link Capabilities
{ 0x88, 2, "LCTL" }, // Link Control
{ 0x8A, 2, "LSTS" }, // Link Status
/* { 0x8C, 4, "RSVD" }, // Reserved
{ 0x90, 4, "RSVD" }, // Reserved
{ 0x94, 4, "RSVD" }, // Reserved */
{ 0x98, 2, "LCTL2" }, // Link Control 2
{ 0x9A, 2, "LSTS2" }, // Link Status 2
/* ... - Reserved */
{ 0xBC0, 4, "AFE_BMUF0" }, // AFE BMU Configuration Function 0
{ 0xBC4, 4, "RSVD" }, // Reserved
{ 0xBC8, 4, "RSVD" }, // Reserved
{ 0xBCC, 4, "AFE_BMUT0" }, // AFE BMU Configuration Test 0
/* ... - Reserved */
};
/*
* All Haswell DMI Registers per
*
* Mobile 4th Generation Intel Core TM Processor Family, Mobile Intel Pentium Processor Family,
* and Mobile Intel Celeron Processor Family
* Datasheet Volume 2
* 329002-002
*/
static const io_register_t haswell_ult_dmi_registers[] = {
{ 0x00, 4, "DMIVCECH" }, // DMI Virtual Channel Enhanced Capability
{ 0x04, 4, "DMIPVCCAP1" }, // DMI Port VC Capability Register 1
{ 0x08, 4, "DMIPVCCAP2" }, // DMI Port VC Capability Register 2
{ 0x0C, 2, "DMI PVCCTL" }, // DMI Port VC Control
/* { 0x0E, 2, "RSVD" }, // Reserved */
{ 0x10, 4, "DMIVC0RCAP" }, // DMI VC0 Resource Capability
{ 0x14, 4, "DMIVC0RCTL" }, // DMI VC0 Resource Control
/* { 0x18, 2, "RSVD" }, // Reserved */
{ 0x1A, 2, "DMIVC0RSTS" }, // DMI VC0 Resource Status
{ 0x1C, 4, "DMIVC1RCAP" }, // DMI VC1 Resource Capability
{ 0x20, 4, "DMIVC1RCTL" }, // DMI VC1 Resource Control
/* { 0x24, 2, "RSVD" }, // Reserved */
{ 0x26, 2, "DMIVC1RSTS" }, // DMI VC1 Resource Status
{ 0x28, 4, "DMIVCPRCAP" }, // DMI VCp Resource Capability
{ 0x2C, 4, "DMIVCPRCTL" }, // DMI VCp Resource Control
/* { 0x30, 2, "RSVD" }, // Reserved */
{ 0x32, 2, "DMIVCPRSTS" }, // DMI VCp Resource Status
{ 0x34, 4, "DMIVCMRCAP" }, // DMI VCm Resource Capability
{ 0x38, 4, "DMIVCMRCTL" }, // DMI VCm Resource Control
/* { 0x3C, 2, "RSVD" }, // Reserved */
{ 0x3E, 2, "DMIVCMRSTS" }, // DMI VCm Resource Status
{ 0x40, 4, "DMIRCLDECH" }, // DMI Root Complex Link Declaration */
{ 0x44, 4, "DMIESD" }, // DMI Element Self Description
/* { 0x48, 4, "RSVD" }, // Reserved */
/* { 0x4C, 4, "RSVD" }, // Reserved */
{ 0x50, 4, "DMILE1D" }, // DMI Link Entry 1 Description
/* { 0x54, 4, "RSVD" }, // Reserved */
{ 0x58, 4, "DMILE1A" }, // DMI Link Entry 1 Address
{ 0x5C, 4, "DMILUE1A" }, // DMI Link Upper Entry 1 Address
{ 0x60, 4, "DMILE2D" }, // DMI Link Entry 2 Description
/* { 0x64, 4, "RSVD" }, // Reserved */
{ 0x68, 4, "DMILE2A" }, // DMI Link Entry 2 Address
/* { 0x6C, 4, "RSVD" }, // Reserved */
/* { 0x70, 4, "RSVD" }, // Reserved */
/* { 0x74, 4, "RSVD" }, // Reserved */
/* { 0x78, 4, "RSVD" }, // Reserved */
/* { 0x7C, 4, "RSVD" }, // Reserved */
/* { 0x80, 4, "RSVD" }, // Reserved */
/* { 0x84, 4, "RSVD" }, // Reserved */
{ 0x88, 2, "LCTL" }, // Link Control
/* ... - Reserved */
{ 0x1C4, 4, "DMIUESTS" }, // DMI Uncorrectable Error Status
{ 0x1C8, 4, "DMIUEMSK" }, // DMI Uncorrectable Error Mask
{ 0x1D0, 4, "DMICESTS" }, // DMI Correctable Error Status
{ 0x1D4, 4, "DMICEMSK" }, // DMI Correctable Error Mask
/* ... - Reserved */
};
/*
* Egress Port Root Complex MMIO configuration space
*/
int print_epbar(struct pci_dev *nb)
{
int i, size = (4 * 1024);
volatile uint8_t *epbar;
uint64_t epbar_phys;
printf("\n============= EPBAR =============\n\n");
switch (nb->device_id) {
case PCI_DEVICE_ID_INTEL_82915:
case PCI_DEVICE_ID_INTEL_82945GM:
case PCI_DEVICE_ID_INTEL_82945GSE:
case PCI_DEVICE_ID_INTEL_82945P:
case PCI_DEVICE_ID_INTEL_82946:
case PCI_DEVICE_ID_INTEL_82975X:
epbar_phys = pci_read_long(nb, 0x40) & 0xfffffffe;
break;
case PCI_DEVICE_ID_INTEL_82965PM:
case PCI_DEVICE_ID_INTEL_82Q965:
case PCI_DEVICE_ID_INTEL_82Q35:
case PCI_DEVICE_ID_INTEL_82G33:
case PCI_DEVICE_ID_INTEL_82Q33:
case PCI_DEVICE_ID_INTEL_82X38:
case PCI_DEVICE_ID_INTEL_32X0:
case PCI_DEVICE_ID_INTEL_82X4X:
case PCI_DEVICE_ID_INTEL_ATOM_DXXX:
case PCI_DEVICE_ID_INTEL_ATOM_NXXX:
case PCI_DEVICE_ID_INTEL_CORE_4TH_GEN_U:
epbar_phys = pci_read_long(nb, 0x40) & 0xfffffffe;
epbar_phys |= ((uint64_t)pci_read_long(nb, 0x44)) << 32;
break;
case PCI_DEVICE_ID_INTEL_82810:
case PCI_DEVICE_ID_INTEL_82810_DC:
case PCI_DEVICE_ID_INTEL_82810E_DC:
case PCI_DEVICE_ID_INTEL_82830M:
case PCI_DEVICE_ID_INTEL_82865:
printf("This northbridge does not have EPBAR.\n");
return 1;
default:
printf("Error: Dumping EPBAR on this northbridge is not (yet) supported.\n");
return 1;
}
epbar = map_physical(epbar_phys, size);
if (epbar == NULL) {
perror("Error mapping EPBAR");
exit(1);
}
printf("EPBAR = 0x%08" PRIx64 " (MEM)\n\n", epbar_phys);
for (i = 0; i < size; i += 4) {
if (*(uint32_t *)(epbar + i))
printf("0x%04x: 0x%08x\n", i, *(uint32_t *)(epbar+i));
}
unmap_physical((void *)epbar, size);
return 0;
}
/*
* MCH-ICH Serial Interconnect Ingress Root Complex MMIO configuration space
*/
int print_dmibar(struct pci_dev *nb)
{
int i, size = (4 * 1024);
volatile uint8_t *dmibar;
uint64_t dmibar_phys;
const io_register_t *dmi_registers = NULL;
printf("\n============= DMIBAR ============\n\n");
switch (nb->device_id) {
case PCI_DEVICE_ID_INTEL_82915:
case PCI_DEVICE_ID_INTEL_82945GM:
case PCI_DEVICE_ID_INTEL_82945GSE:
case PCI_DEVICE_ID_INTEL_82945P:
case PCI_DEVICE_ID_INTEL_82975X:
dmibar_phys = pci_read_long(nb, 0x4c) & 0xfffffffe;
break;
case PCI_DEVICE_ID_INTEL_82946:
case PCI_DEVICE_ID_INTEL_82965PM:
case PCI_DEVICE_ID_INTEL_82Q965:
case PCI_DEVICE_ID_INTEL_82Q35:
case PCI_DEVICE_ID_INTEL_82G33:
case PCI_DEVICE_ID_INTEL_82Q33:
case PCI_DEVICE_ID_INTEL_82X38:
case PCI_DEVICE_ID_INTEL_32X0:
case PCI_DEVICE_ID_INTEL_82X4X:
case PCI_DEVICE_ID_INTEL_ATOM_DXXX:
case PCI_DEVICE_ID_INTEL_ATOM_NXXX:
dmibar_phys = pci_read_long(nb, 0x68) & 0xfffffffe;
dmibar_phys |= ((uint64_t)pci_read_long(nb, 0x6c)) << 32;
break;
case PCI_DEVICE_ID_INTEL_82810:
case PCI_DEVICE_ID_INTEL_82810_DC:
case PCI_DEVICE_ID_INTEL_82810E_DC:
case PCI_DEVICE_ID_INTEL_82865:
printf("This northbridge does not have DMIBAR.\n");
return 1;
case PCI_DEVICE_ID_INTEL_82X58:
dmibar_phys = pci_read_long(nb, 0x50) & 0xfffff000;
break;
case PCI_DEVICE_ID_INTEL_CORE_0TH_GEN:
/* DMIBAR is called DMIRCBAR in Nehalem */
dmibar_phys = pci_read_long(nb, 0x50) & 0xfffff000; /* 31:12 */
dmi_registers = nehalem_dmi_registers;
size = ARRAY_SIZE(nehalem_dmi_registers);
break;
case PCI_DEVICE_ID_INTEL_CORE_1ST_GEN:
dmibar_phys = pci_read_long(nb, 0x68);
dmibar_phys |= ((uint64_t)pci_read_long(nb, 0x6c)) << 32;
dmibar_phys &= 0x0000000ffffff000UL; /* 35:12 */
dmi_registers = westmere_dmi_registers;
size = ARRAY_SIZE(westmere_dmi_registers);
break;
case PCI_DEVICE_ID_INTEL_CORE_2ND_GEN:
dmi_registers = sandybridge_dmi_registers;
size = ARRAY_SIZE(sandybridge_dmi_registers);
case PCI_DEVICE_ID_INTEL_CORE_3RD_GEN_A: /* pretty printing not implemented yet */
case PCI_DEVICE_ID_INTEL_CORE_3RD_GEN_B:
case PCI_DEVICE_ID_INTEL_CORE_3RD_GEN_C:
case PCI_DEVICE_ID_INTEL_CORE_3RD_GEN_D:
dmibar_phys = pci_read_long(nb, 0x68);
dmibar_phys |= ((uint64_t)pci_read_long(nb, 0x6c)) << 32;
dmibar_phys &= 0x0000007ffffff000UL; /* 38:12 */
break;
case PCI_DEVICE_ID_INTEL_CORE_4TH_GEN_U:
dmi_registers = haswell_ult_dmi_registers;
size = ARRAY_SIZE(haswell_ult_dmi_registers);
dmibar_phys = pci_read_long(nb, 0x68);
dmibar_phys |= ((uint64_t)pci_read_long(nb, 0x6c)) << 32;
dmibar_phys &= 0x0000007ffffff000UL; /* 38:12 */
break;
default:
printf("Error: Dumping DMIBAR on this northbridge is not (yet) supported.\n");
return 1;
}
dmibar = map_physical(dmibar_phys, size);
if (dmibar == NULL) {
perror("Error mapping DMIBAR");
exit(1);
}
printf("DMIBAR = 0x%08" PRIx64 " (MEM)\n\n", dmibar_phys);
if (dmi_registers != NULL) {
for (i = 0; i < size; i++) {
switch (dmi_registers[i].size) {
case 4:
printf("dmibase+0x%04x: 0x%08x (%s)\n",
dmi_registers[i].addr,
*(uint32_t *)(dmibar+dmi_registers[i].addr),
dmi_registers[i].name);
break;
case 2:
printf("dmibase+0x%04x: 0x%04x (%s)\n",
dmi_registers[i].addr,
*(uint16_t *)(dmibar+dmi_registers[i].addr),
dmi_registers[i].name);
break;
case 1:
printf("dmibase+0x%04x: 0x%02x (%s)\n",
dmi_registers[i].addr,
*(uint8_t *)(dmibar+dmi_registers[i].addr),
dmi_registers[i].name);
break;
}
}
} else {
for (i = 0; i < size; i += 4) {
if (*(uint32_t *)(dmibar + i))
printf("0x%04x: 0x%08x\n", i, *(uint32_t *)(dmibar+i));
}
}
unmap_physical((void *)dmibar, size);
return 0;
}
/*
* PCIe MMIO configuration space
*/
int print_pciexbar(struct pci_dev *nb)
{
uint64_t pciexbar_reg;
uint64_t pciexbar_phys;
volatile uint8_t *pciexbar;
int max_busses, devbase, i;
int bus, dev, fn;
printf("========= PCIEXBAR ========\n\n");
switch (nb->device_id) {
case PCI_DEVICE_ID_INTEL_82915:
case PCI_DEVICE_ID_INTEL_82945GM:
case PCI_DEVICE_ID_INTEL_82945GSE:
case PCI_DEVICE_ID_INTEL_82945P:
case PCI_DEVICE_ID_INTEL_82975X:
pciexbar_reg = pci_read_long(nb, 0x48);
break;
case PCI_DEVICE_ID_INTEL_82946:
case PCI_DEVICE_ID_INTEL_82965PM:
case PCI_DEVICE_ID_INTEL_82Q965:
case PCI_DEVICE_ID_INTEL_82Q35:
case PCI_DEVICE_ID_INTEL_82G33:
case PCI_DEVICE_ID_INTEL_82Q33:
case PCI_DEVICE_ID_INTEL_82X38:
case PCI_DEVICE_ID_INTEL_32X0:
case PCI_DEVICE_ID_INTEL_82X4X:
case PCI_DEVICE_ID_INTEL_ATOM_DXXX:
case PCI_DEVICE_ID_INTEL_ATOM_NXXX:
case PCI_DEVICE_ID_INTEL_CORE_4TH_GEN_U:
pciexbar_reg = pci_read_long(nb, 0x60);
pciexbar_reg |= ((uint64_t)pci_read_long(nb, 0x64)) << 32;
break;
case PCI_DEVICE_ID_INTEL_82810:
case PCI_DEVICE_ID_INTEL_82810_DC:
case PCI_DEVICE_ID_INTEL_82810E_DC:
case PCI_DEVICE_ID_INTEL_82865:
printf("Error: This northbridge does not have PCIEXBAR.\n");
return 1;
default:
printf("Error: Dumping PCIEXBAR on this northbridge is not (yet) supported.\n");
return 1;
}
if (!(pciexbar_reg & (1 << 0))) {
printf("PCIEXBAR register is disabled.\n");
return 0;
}
switch ((pciexbar_reg >> 1) & 3) {
case 0: // 256MB
inteltool: pcie.c: Use `0xffULL` instead of `0xff` to avoid shift overflow When building inteltool with Clang, it warns about the following. $ clang --version Debian clang version 3.2-1~exp6 (tags/RELEASE_32/final) (based on LLVM 3.2) Target: i386-pc-linux-gnu Thread model: posix $ CC=clang make […] clang -O2 -g -Wall -W -c -o pcie.o pcie.c pcie.c:297:40: warning: signed shift result (0xFF0000000) requires 37 bits to represent, but 'int' only has 32 bits [-Wshift-overflow] pciexbar_phys = pciexbar_reg & (0xff << 28); ~~~~ ^ ~~ pcie.c:301:41: warning: signed shift result (0xFF8000000) requires 37 bits to represent, but 'int' only has 32 bits [-Wshift-overflow] pciexbar_phys = pciexbar_reg & (0x1ff << 27); ~~~~~ ^ ~~ pcie.c:305:41: warning: signed shift result (0xFFC000000) requires 37 bits to represent, but 'int' only has 32 bits [-Wshift-overflow] pciexbar_phys = pciexbar_reg & (0x3ff << 26); ~~~~~ ^ ~~ 3 warnings generated. […] Specifying the length by using the suffix `0xffULL` fixes these issues as now enough bits are available. These issues were introduced in commit 1162f25a [1]. commit 1162f25a49e8f39822123d664cda10fef466b351 Author: Stefan Reinauer <stepan@coresystems.de> Date: Thu Dec 4 15:18:20 2008 +0000 Patch to util/inteltool: * PMBASE dumping now knows the registers. * Add support for i965, i975, ICH8M * Add support for Darwin OS using DirectIO [1] http://review.coreboot.org/gitweb?p=coreboot.git;a=commit;h=1162f25a49e8f39822123d664cda10fef466b351 Change-Id: I7b9a15b04ef3bcae64e06266667597d0f9f07b79 Signed-off-by: Paul Menzel <paulepanter@users.sourceforge.net> Reviewed-on: http://review.coreboot.org/3015 Tested-by: build bot (Jenkins) Reviewed-by: Nico Huber <nico.huber@secunet.com> Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
2013-04-03 10:00:33 +02:00
pciexbar_phys = pciexbar_reg & (0xffULL << 28);
max_busses = 256;
break;
case 1: // 128M
inteltool: pcie.c: Use `0xffULL` instead of `0xff` to avoid shift overflow When building inteltool with Clang, it warns about the following. $ clang --version Debian clang version 3.2-1~exp6 (tags/RELEASE_32/final) (based on LLVM 3.2) Target: i386-pc-linux-gnu Thread model: posix $ CC=clang make […] clang -O2 -g -Wall -W -c -o pcie.o pcie.c pcie.c:297:40: warning: signed shift result (0xFF0000000) requires 37 bits to represent, but 'int' only has 32 bits [-Wshift-overflow] pciexbar_phys = pciexbar_reg & (0xff << 28); ~~~~ ^ ~~ pcie.c:301:41: warning: signed shift result (0xFF8000000) requires 37 bits to represent, but 'int' only has 32 bits [-Wshift-overflow] pciexbar_phys = pciexbar_reg & (0x1ff << 27); ~~~~~ ^ ~~ pcie.c:305:41: warning: signed shift result (0xFFC000000) requires 37 bits to represent, but 'int' only has 32 bits [-Wshift-overflow] pciexbar_phys = pciexbar_reg & (0x3ff << 26); ~~~~~ ^ ~~ 3 warnings generated. […] Specifying the length by using the suffix `0xffULL` fixes these issues as now enough bits are available. These issues were introduced in commit 1162f25a [1]. commit 1162f25a49e8f39822123d664cda10fef466b351 Author: Stefan Reinauer <stepan@coresystems.de> Date: Thu Dec 4 15:18:20 2008 +0000 Patch to util/inteltool: * PMBASE dumping now knows the registers. * Add support for i965, i975, ICH8M * Add support for Darwin OS using DirectIO [1] http://review.coreboot.org/gitweb?p=coreboot.git;a=commit;h=1162f25a49e8f39822123d664cda10fef466b351 Change-Id: I7b9a15b04ef3bcae64e06266667597d0f9f07b79 Signed-off-by: Paul Menzel <paulepanter@users.sourceforge.net> Reviewed-on: http://review.coreboot.org/3015 Tested-by: build bot (Jenkins) Reviewed-by: Nico Huber <nico.huber@secunet.com> Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
2013-04-03 10:00:33 +02:00
pciexbar_phys = pciexbar_reg & (0x1ffULL << 27);
max_busses = 128;
break;
case 2: // 64M
inteltool: pcie.c: Use `0xffULL` instead of `0xff` to avoid shift overflow When building inteltool with Clang, it warns about the following. $ clang --version Debian clang version 3.2-1~exp6 (tags/RELEASE_32/final) (based on LLVM 3.2) Target: i386-pc-linux-gnu Thread model: posix $ CC=clang make […] clang -O2 -g -Wall -W -c -o pcie.o pcie.c pcie.c:297:40: warning: signed shift result (0xFF0000000) requires 37 bits to represent, but 'int' only has 32 bits [-Wshift-overflow] pciexbar_phys = pciexbar_reg & (0xff << 28); ~~~~ ^ ~~ pcie.c:301:41: warning: signed shift result (0xFF8000000) requires 37 bits to represent, but 'int' only has 32 bits [-Wshift-overflow] pciexbar_phys = pciexbar_reg & (0x1ff << 27); ~~~~~ ^ ~~ pcie.c:305:41: warning: signed shift result (0xFFC000000) requires 37 bits to represent, but 'int' only has 32 bits [-Wshift-overflow] pciexbar_phys = pciexbar_reg & (0x3ff << 26); ~~~~~ ^ ~~ 3 warnings generated. […] Specifying the length by using the suffix `0xffULL` fixes these issues as now enough bits are available. These issues were introduced in commit 1162f25a [1]. commit 1162f25a49e8f39822123d664cda10fef466b351 Author: Stefan Reinauer <stepan@coresystems.de> Date: Thu Dec 4 15:18:20 2008 +0000 Patch to util/inteltool: * PMBASE dumping now knows the registers. * Add support for i965, i975, ICH8M * Add support for Darwin OS using DirectIO [1] http://review.coreboot.org/gitweb?p=coreboot.git;a=commit;h=1162f25a49e8f39822123d664cda10fef466b351 Change-Id: I7b9a15b04ef3bcae64e06266667597d0f9f07b79 Signed-off-by: Paul Menzel <paulepanter@users.sourceforge.net> Reviewed-on: http://review.coreboot.org/3015 Tested-by: build bot (Jenkins) Reviewed-by: Nico Huber <nico.huber@secunet.com> Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
2013-04-03 10:00:33 +02:00
pciexbar_phys = pciexbar_reg & (0x3ffULL << 26);
max_busses = 64;
break;
default: // RSVD
printf("Undefined address base. Bailing out.\n");
return 1;
}
printf("PCIEXBAR: 0x%08" PRIx64 "\n", pciexbar_phys);
pciexbar = map_physical(pciexbar_phys, (max_busses * 1024 * 1024));
if (pciexbar == NULL) {
perror("Error mapping PCIEXBAR");
exit(1);
}
for (bus = 0; bus < max_busses; bus++) {
for (dev = 0; dev < 32; dev++) {
for (fn = 0; fn < 8; fn++) {
devbase = (bus * 1024 * 1024) + (dev * 32 * 1024) + (fn * 4 * 1024);
if (*(uint16_t *)(pciexbar + devbase) == 0xffff)
continue;
/* This is a heuristics. Anyone got a better check? */
if( (*(uint32_t *)(pciexbar + devbase + 256) == 0xffffffff) &&
(*(uint32_t *)(pciexbar + devbase + 512) == 0xffffffff) ) {
#if DEBUG
printf("Skipped non-PCIe device %02x:%02x.%01x\n", bus, dev, fn);
#endif
continue;
}
printf("\nPCIe %02x:%02x.%01x extended config space:", bus, dev, fn);
for (i = 0; i < 4096; i++) {
if((i % 0x10) == 0)
printf("\n%04x:", i);
printf(" %02x", *(pciexbar+devbase+i));
}
printf("\n");
}
}
}
unmap_physical((void *)pciexbar, (max_busses * 1024 * 1024));
return 0;
}