linux/arch/x86/pci/acpi.c

473 lines
12 KiB
C
Raw Normal View History

#include <linux/pci.h>
#include <linux/acpi.h>
#include <linux/init.h>
#include <linux/irq.h>
#include <linux/dmi.h>
include cleanup: Update gfp.h and slab.h includes to prepare for breaking implicit slab.h inclusion from percpu.h percpu.h is included by sched.h and module.h and thus ends up being included when building most .c files. percpu.h includes slab.h which in turn includes gfp.h making everything defined by the two files universally available and complicating inclusion dependencies. percpu.h -> slab.h dependency is about to be removed. Prepare for this change by updating users of gfp and slab facilities include those headers directly instead of assuming availability. As this conversion needs to touch large number of source files, the following script is used as the basis of conversion. http://userweb.kernel.org/~tj/misc/slabh-sweep.py The script does the followings. * Scan files for gfp and slab usages and update includes such that only the necessary includes are there. ie. if only gfp is used, gfp.h, if slab is used, slab.h. * When the script inserts a new include, it looks at the include blocks and try to put the new include such that its order conforms to its surrounding. It's put in the include block which contains core kernel includes, in the same order that the rest are ordered - alphabetical, Christmas tree, rev-Xmas-tree or at the end if there doesn't seem to be any matching order. * If the script can't find a place to put a new include (mostly because the file doesn't have fitting include block), it prints out an error message indicating which .h file needs to be added to the file. The conversion was done in the following steps. 1. The initial automatic conversion of all .c files updated slightly over 4000 files, deleting around 700 includes and adding ~480 gfp.h and ~3000 slab.h inclusions. The script emitted errors for ~400 files. 2. Each error was manually checked. Some didn't need the inclusion, some needed manual addition while adding it to implementation .h or embedding .c file was more appropriate for others. This step added inclusions to around 150 files. 3. The script was run again and the output was compared to the edits from #2 to make sure no file was left behind. 4. Several build tests were done and a couple of problems were fixed. e.g. lib/decompress_*.c used malloc/free() wrappers around slab APIs requiring slab.h to be added manually. 5. The script was run on all .h files but without automatically editing them as sprinkling gfp.h and slab.h inclusions around .h files could easily lead to inclusion dependency hell. Most gfp.h inclusion directives were ignored as stuff from gfp.h was usually wildly available and often used in preprocessor macros. Each slab.h inclusion directive was examined and added manually as necessary. 6. percpu.h was updated not to include slab.h. 7. Build test were done on the following configurations and failures were fixed. CONFIG_GCOV_KERNEL was turned off for all tests (as my distributed build env didn't work with gcov compiles) and a few more options had to be turned off depending on archs to make things build (like ipr on powerpc/64 which failed due to missing writeq). * x86 and x86_64 UP and SMP allmodconfig and a custom test config. * powerpc and powerpc64 SMP allmodconfig * sparc and sparc64 SMP allmodconfig * ia64 SMP allmodconfig * s390 SMP allmodconfig * alpha SMP allmodconfig * um on x86_64 SMP allmodconfig 8. percpu.h modifications were reverted so that it could be applied as a separate patch and serve as bisection point. Given the fact that I had only a couple of failures from tests on step 6, I'm fairly confident about the coverage of this conversion patch. If there is a breakage, it's likely to be something in one of the arch headers which should be easily discoverable easily on most builds of the specific arch. Signed-off-by: Tejun Heo <tj@kernel.org> Guess-its-ok-by: Christoph Lameter <cl@linux-foundation.org> Cc: Ingo Molnar <mingo@redhat.com> Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
2010-03-24 08:04:11 +00:00
#include <linux/slab.h>
#include <asm/numa.h>
#include <asm/pci_x86.h>
struct pci_root_info {
struct acpi_device *bridge;
char *name;
unsigned int res_num;
struct resource *res;
struct pci_bus *bus;
int busnum;
};
static bool pci_use_crs = true;
static int __init set_use_crs(const struct dmi_system_id *id)
{
pci_use_crs = true;
return 0;
}
static int __init set_nouse_crs(const struct dmi_system_id *id)
{
pci_use_crs = false;
return 0;
}
static const struct dmi_system_id pci_use_crs_table[] __initconst = {
/* http://bugzilla.kernel.org/show_bug.cgi?id=14183 */
{
.callback = set_use_crs,
.ident = "IBM System x3800",
.matches = {
DMI_MATCH(DMI_SYS_VENDOR, "IBM"),
DMI_MATCH(DMI_PRODUCT_NAME, "x3800"),
},
},
x86/PCI: use host bridge _CRS info on ASRock ALiveSATA2-GLAN This DMI quirk turns on "pci=use_crs" for the ALiveSATA2-GLAN because amd_bus.c doesn't handle this system correctly. The system has a single HyperTransport I/O chain, but has two PCI host bridges to buses 00 and 80. amd_bus.c learns the MMIO range associated with buses 00-ff and that this range is routed to the HT chain hosted at node 0, link 0: bus: [00, ff] on node 0 link 0 bus: 00 index 1 [mem 0x80000000-0xfcffffffff] This includes the address space for both bus 00 and bus 80, and amd_bus.c assumes it's all routed to bus 00. We find device 80:01.0, which BIOS left in the middle of that space, but we don't find a bridge from bus 00 to bus 80, so we conclude that 80:01.0 is unreachable from bus 00, and we move it from the original, working, address to something outside the bus 00 aperture, which does not work: pci 0000:80:01.0: reg 10: [mem 0xfebfc000-0xfebfffff 64bit] pci 0000:80:01.0: BAR 0: assigned [mem 0xfd00000000-0xfd00003fff 64bit] The BIOS told us everything we need to know to handle this correctly, so we're better off if we just pay attention, which lets us leave the 80:01.0 device at the original, working, address: ACPI: PCI Root Bridge [PCI0] (domain 0000 [bus 00-7f]) pci_root PNP0A03:00: host bridge window [mem 0x80000000-0xff37ffff] ACPI: PCI Root Bridge [PCI1] (domain 0000 [bus 80-ff]) pci_root PNP0A08:00: host bridge window [mem 0xfebfc000-0xfebfffff] This was a regression between 2.6.33 and 2.6.34. In 2.6.33, amd_bus.c was used only when we found multiple HT chains. 3e3da00c01d050, which enabled amd_bus.c even on systems with a single HT chain, caused this failure. This quirk was written by Graham. If we ever enable "pci=use_crs" for machines from 2006 or earlir, this quirk should be removed. Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=16007 Cc: stable@kernel.org Reported-by: Graham Ramsey <ramsey.graham@ntlworld.com> Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com> Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
2010-07-23 18:53:27 +00:00
/* https://bugzilla.kernel.org/show_bug.cgi?id=16007 */
/* 2006 AMD HT/VIA system with two host bridges */
{
.callback = set_use_crs,
.ident = "ASRock ALiveSATA2-GLAN",
.matches = {
DMI_MATCH(DMI_PRODUCT_NAME, "ALiveSATA2-GLAN"),
},
},
x86/PCI: use host bridge _CRS info on ASUS M2V-MX SE In summary, this DMI quirk uses the _CRS info by default for the ASUS M2V-MX SE by turning on `pci=use_crs` and is similar to the quirk added by commit 2491762cfb47 ("x86/PCI: use host bridge _CRS info on ASRock ALiveSATA2-GLAN") whose commit message should be read for further information. Since commit 3e3da00c01d0 ("x86/pci: AMD one chain system to use pci read out res") Linux gives the following oops: parport0: PC-style at 0x378, irq 7 [PCSPP,TRISTATE] HDA Intel 0000:20:01.0: PCI INT A -> GSI 17 (level, low) -> IRQ 17 HDA Intel 0000:20:01.0: setting latency timer to 64 BUG: unable to handle kernel paging request at ffffc90011c08000 IP: [<ffffffffa0578402>] azx_probe+0x3ad/0x86b [snd_hda_intel] PGD 13781a067 PUD 13781b067 PMD 1300ba067 PTE 800000fd00000173 Oops: 0009 [#1] SMP last sysfs file: /sys/module/snd_pcm/initstate CPU 0 Modules linked in: snd_hda_intel(+) snd_hda_codec snd_hwdep snd_pcm_oss snd_mixer_oss snd_pcm snd_seq_midi snd_rawmidi snd_seq_midi_event tpm_tis tpm snd_seq tpm_bios psmouse parport_pc snd_timer snd_seq_device parport processor evdev snd i2c_viapro thermal_sys amd64_edac_mod k8temp i2c_core soundcore shpchp pcspkr serio_raw asus_atk0110 pci_hotplug edac_core button snd_page_alloc edac_mce_amd ext3 jbd mbcache sha256_generic cryptd aes_x86_64 aes_generic cbc dm_crypt dm_mod raid1 md_mod usbhid hid sg sd_mod crc_t10dif sr_mod cdrom ata_generic uhci_hcd sata_via pata_via libata ehci_hcd usbcore scsi_mod via_rhine mii nls_base [last unloaded: scsi_wait_scan] Pid: 1153, comm: work_for_cpu Not tainted 2.6.37-1-amd64 #1 M2V-MX SE/System Product Name RIP: 0010:[<ffffffffa0578402>] [<ffffffffa0578402>] azx_probe+0x3ad/0x86b [snd_hda_intel] RSP: 0018:ffff88013153fe50 EFLAGS: 00010286 RAX: ffffc90011c08000 RBX: ffff88013029ec00 RCX: 0000000000000006 RDX: 0000000000000000 RSI: 0000000000000246 RDI: 0000000000000246 RBP: ffff88013341d000 R08: 0000000000000000 R09: 0000000000000040 R10: 0000000000000286 R11: 0000000000003731 R12: ffff88013029c400 R13: 0000000000000000 R14: 0000000000000000 R15: ffff88013341d090 FS: 0000000000000000(0000) GS:ffff8800bfc00000(0000) knlGS:00000000f7610ab0 CS: 0010 DS: 0000 ES: 0000 CR0: 000000008005003b CR2: ffffc90011c08000 CR3: 0000000132f57000 CR4: 00000000000006f0 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400 Process work_for_cpu (pid: 1153, threadinfo ffff88013153e000, task ffff8801303c86c0) Stack: 0000000000000005 ffffffff8123ad65 00000000000136c0 ffff88013029c400 ffff8801303c8998 ffff88013341d000 ffff88013341d090 ffff8801322d9dc8 ffff88013341d208 0000000000000000 0000000000000000 ffffffff811ad232 Call Trace: [<ffffffff8123ad65>] ? __pm_runtime_set_status+0x162/0x186 [<ffffffff811ad232>] ? local_pci_probe+0x49/0x92 [<ffffffff8105afc5>] ? do_work_for_cpu+0x0/0x1b [<ffffffff8105afc5>] ? do_work_for_cpu+0x0/0x1b [<ffffffff8105afd0>] ? do_work_for_cpu+0xb/0x1b [<ffffffff8105fd3f>] ? kthread+0x7a/0x82 [<ffffffff8100a824>] ? kernel_thread_helper+0x4/0x10 [<ffffffff8105fcc5>] ? kthread+0x0/0x82 [<ffffffff8100a820>] ? kernel_thread_helper+0x0/0x10 Code: f4 01 00 00 ef 31 f6 48 89 df e8 29 dd ff ff 85 c0 0f 88 2b 03 00 00 48 89 ef e8 b4 39 c3 e0 8b 7b 40 e8 fc 9d b1 e0 48 8b 43 38 <66> 8b 10 66 89 14 24 8b 43 14 83 e8 03 83 f8 01 77 32 31 d2 be RIP [<ffffffffa0578402>] azx_probe+0x3ad/0x86b [snd_hda_intel] RSP <ffff88013153fe50> CR2: ffffc90011c08000 ---[ end trace 8d1f3ebc136437fd ]--- Trusting the ACPI _CRS information (`pci=use_crs`) fixes this problem. $ dmesg | grep -i crs # with the quirk PCI: Using host bridge windows from ACPI; if necessary, use "pci=nocrs" and report a bug The match has to be against the DMI board entries though since the vendor entries are not populated. DMI: System manufacturer System Product Name/M2V-MX SE, BIOS 0304 10/30/2007 This quirk should be removed when `pci=use_crs` is enabled for machines from 2006 or earlier or some other solution is implemented. Using coreboot [1] with this board the problem does not exist but this quirk also does not affect it either. To be safe though the check is tightened to only take effect when the BIOS from American Megatrends is used. 15:13 < ruik> but coreboot does not need that 15:13 < ruik> because i have there only one root bus 15:13 < ruik> the audio is behind a bridge $ sudo dmidecode BIOS Information Vendor: American Megatrends Inc. Version: 0304 Release Date: 10/30/2007 [1] http://www.coreboot.org/ Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=30552 Cc: stable@kernel.org (2.6.34) Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Ingo Molnar <mingo@redhat.com> Cc: H. Peter Anvin <hpa@zytor.com> Cc: x86@kernel.org Signed-off-by: Paul Menzel <paulepanter@users.sourceforge.net> Signed-off-by: Bjorn Helgaas <bhelgaas@google.com> Acked-by: Jesse Barnes <jbarnes@virtuousgeek.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2011-08-31 15:07:10 +00:00
/* https://bugzilla.kernel.org/show_bug.cgi?id=30552 */
/* 2006 AMD HT/VIA system with two host bridges */
{
.callback = set_use_crs,
.ident = "ASUS M2V-MX SE",
.matches = {
DMI_MATCH(DMI_BOARD_VENDOR, "ASUSTeK Computer INC."),
DMI_MATCH(DMI_BOARD_NAME, "M2V-MX SE"),
DMI_MATCH(DMI_BIOS_VENDOR, "American Megatrends Inc."),
},
},
/* Now for the blacklist.. */
/* https://bugzilla.redhat.com/show_bug.cgi?id=769657 */
{
.callback = set_nouse_crs,
.ident = "Dell Studio 1557",
.matches = {
DMI_MATCH(DMI_BOARD_VENDOR, "Dell Inc."),
DMI_MATCH(DMI_PRODUCT_NAME, "Studio 1557"),
DMI_MATCH(DMI_BIOS_VERSION, "A09"),
},
},
/* https://bugzilla.redhat.com/show_bug.cgi?id=769657 */
{
.callback = set_nouse_crs,
.ident = "Thinkpad SL510",
.matches = {
DMI_MATCH(DMI_BOARD_VENDOR, "LENOVO"),
DMI_MATCH(DMI_BOARD_NAME, "2847DFG"),
DMI_MATCH(DMI_BIOS_VERSION, "6JET85WW (1.43 )"),
},
},
{}
};
void __init pci_acpi_crs_quirks(void)
{
int year;
if (dmi_get_date(DMI_BIOS_DATE, &year, NULL, NULL) && year < 2008)
pci_use_crs = false;
dmi_check_system(pci_use_crs_table);
/*
* If the user specifies "pci=use_crs" or "pci=nocrs" explicitly, that
* takes precedence over anything we figured out above.
*/
if (pci_probe & PCI_ROOT_NO_CRS)
pci_use_crs = false;
else if (pci_probe & PCI_USE__CRS)
pci_use_crs = true;
printk(KERN_INFO "PCI: %s host bridge windows from ACPI; "
"if necessary, use \"pci=%s\" and report a bug\n",
pci_use_crs ? "Using" : "Ignoring",
pci_use_crs ? "nocrs" : "use_crs");
}
static acpi_status
resource_to_addr(struct acpi_resource *resource,
struct acpi_resource_address64 *addr)
{
acpi_status status;
struct acpi_resource_memory24 *memory24;
struct acpi_resource_memory32 *memory32;
struct acpi_resource_fixed_memory32 *fixed_memory32;
memset(addr, 0, sizeof(*addr));
switch (resource->type) {
case ACPI_RESOURCE_TYPE_MEMORY24:
memory24 = &resource->data.memory24;
addr->resource_type = ACPI_MEMORY_RANGE;
addr->minimum = memory24->minimum;
addr->address_length = memory24->address_length;
addr->maximum = addr->minimum + addr->address_length - 1;
return AE_OK;
case ACPI_RESOURCE_TYPE_MEMORY32:
memory32 = &resource->data.memory32;
addr->resource_type = ACPI_MEMORY_RANGE;
addr->minimum = memory32->minimum;
addr->address_length = memory32->address_length;
addr->maximum = addr->minimum + addr->address_length - 1;
return AE_OK;
case ACPI_RESOURCE_TYPE_FIXED_MEMORY32:
fixed_memory32 = &resource->data.fixed_memory32;
addr->resource_type = ACPI_MEMORY_RANGE;
addr->minimum = fixed_memory32->address;
addr->address_length = fixed_memory32->address_length;
addr->maximum = addr->minimum + addr->address_length - 1;
return AE_OK;
case ACPI_RESOURCE_TYPE_ADDRESS16:
case ACPI_RESOURCE_TYPE_ADDRESS32:
case ACPI_RESOURCE_TYPE_ADDRESS64:
status = acpi_resource_to_address64(resource, addr);
if (ACPI_SUCCESS(status) &&
(addr->resource_type == ACPI_MEMORY_RANGE ||
addr->resource_type == ACPI_IO_RANGE) &&
addr->address_length > 0) {
return AE_OK;
}
break;
}
return AE_ERROR;
}
static acpi_status
count_resource(struct acpi_resource *acpi_res, void *data)
{
struct pci_root_info *info = data;
struct acpi_resource_address64 addr;
acpi_status status;
status = resource_to_addr(acpi_res, &addr);
if (ACPI_SUCCESS(status))
info->res_num++;
return AE_OK;
}
static acpi_status
setup_resource(struct acpi_resource *acpi_res, void *data)
{
struct pci_root_info *info = data;
struct resource *res;
struct acpi_resource_address64 addr;
acpi_status status;
unsigned long flags;
u64 start, orig_end, end;
status = resource_to_addr(acpi_res, &addr);
if (!ACPI_SUCCESS(status))
return AE_OK;
if (addr.resource_type == ACPI_MEMORY_RANGE) {
flags = IORESOURCE_MEM;
if (addr.info.mem.caching == ACPI_PREFETCHABLE_MEMORY)
flags |= IORESOURCE_PREFETCH;
} else if (addr.resource_type == ACPI_IO_RANGE) {
flags = IORESOURCE_IO;
} else
return AE_OK;
start = addr.minimum + addr.translation_offset;
orig_end = end = addr.maximum + addr.translation_offset;
/* Exclude non-addressable range or non-addressable portion of range */
end = min(end, (u64)iomem_resource.end);
if (end <= start) {
dev_info(&info->bridge->dev,
"host bridge window [%#llx-%#llx] "
"(ignored, not CPU addressable)\n", start, orig_end);
return AE_OK;
} else if (orig_end != end) {
dev_info(&info->bridge->dev,
"host bridge window [%#llx-%#llx] "
"([%#llx-%#llx] ignored, not CPU addressable)\n",
start, orig_end, end + 1, orig_end);
}
res = &info->res[info->res_num];
res->name = info->name;
res->flags = flags;
res->start = start;
res->end = end;
res->child = NULL;
if (!pci_use_crs) {
dev_printk(KERN_DEBUG, &info->bridge->dev,
"host bridge window %pR (ignored)\n", res);
return AE_OK;
}
info->res_num++;
if (addr.translation_offset)
dev_info(&info->bridge->dev, "host bridge window %pR "
"(PCI address [%#llx-%#llx])\n",
res, res->start - addr.translation_offset,
res->end - addr.translation_offset);
else
dev_info(&info->bridge->dev, "host bridge window %pR\n", res);
return AE_OK;
}
static bool resource_contains(struct resource *res, resource_size_t point)
{
if (res->start <= point && point <= res->end)
return true;
return false;
}
static void coalesce_windows(struct pci_root_info *info, unsigned long type)
{
int i, j;
struct resource *res1, *res2;
for (i = 0; i < info->res_num; i++) {
res1 = &info->res[i];
if (!(res1->flags & type))
continue;
for (j = i + 1; j < info->res_num; j++) {
res2 = &info->res[j];
if (!(res2->flags & type))
continue;
/*
* I don't like throwing away windows because then
* our resources no longer match the ACPI _CRS, but
* the kernel resource tree doesn't allow overlaps.
*/
if (resource_contains(res1, res2->start) ||
resource_contains(res1, res2->end) ||
resource_contains(res2, res1->start) ||
resource_contains(res2, res1->end)) {
res1->start = min(res1->start, res2->start);
res1->end = max(res1->end, res2->end);
dev_info(&info->bridge->dev,
"host bridge window expanded to %pR; %pR ignored\n",
res1, res2);
res2->flags = 0;
}
}
}
}
static void add_resources(struct pci_root_info *info)
{
int i;
struct resource *res, *root, *conflict;
if (!pci_use_crs)
return;
coalesce_windows(info, IORESOURCE_MEM);
coalesce_windows(info, IORESOURCE_IO);
for (i = 0; i < info->res_num; i++) {
res = &info->res[i];
if (res->flags & IORESOURCE_MEM)
root = &iomem_resource;
else if (res->flags & IORESOURCE_IO)
root = &ioport_resource;
else
continue;
conflict = insert_resource_conflict(root, res);
if (conflict)
dev_info(&info->bridge->dev,
"ignoring host bridge window %pR (conflicts with %s %pR)\n",
res, conflict->name, conflict);
else
pci_bus_add_resource(info->bus, res, 0);
}
}
static void
get_current_resources(struct acpi_device *device, int busnum,
int domain, struct pci_bus *bus)
{
struct pci_root_info info;
size_t size;
if (pci_use_crs)
pci_bus_remove_resources(bus);
info.bridge = device;
info.bus = bus;
info.res_num = 0;
acpi_walk_resources(device->handle, METHOD_NAME__CRS, count_resource,
&info);
if (!info.res_num)
return;
size = sizeof(*info.res) * info.res_num;
info.res = kmalloc(size, GFP_KERNEL);
if (!info.res)
goto res_alloc_fail;
info.name = kasprintf(GFP_KERNEL, "PCI Bus %04x:%02x", domain, busnum);
if (!info.name)
goto name_alloc_fail;
info.res_num = 0;
acpi_walk_resources(device->handle, METHOD_NAME__CRS, setup_resource,
&info);
add_resources(&info);
return;
name_alloc_fail:
kfree(info.res);
res_alloc_fail:
return;
}
struct pci_bus * __devinit pci_acpi_scan_root(struct acpi_pci_root *root)
{
struct acpi_device *device = root->device;
int domain = root->segment;
int busnum = root->secondary.start;
struct pci_bus *bus;
struct pci_sysdata *sd;
int node;
#ifdef CONFIG_ACPI_NUMA
int pxm;
#endif
if (domain && !pci_domains_supported) {
printk(KERN_WARNING "pci_bus %04x:%02x: "
"ignored (multiple domains not supported)\n",
domain, busnum);
return NULL;
}
node = -1;
#ifdef CONFIG_ACPI_NUMA
pxm = acpi_get_pxm(device->handle);
if (pxm >= 0)
node = pxm_to_node(pxm);
if (node != -1)
set_mp_bus_to_node(busnum, node);
else
#endif
node = get_mp_bus_to_node(busnum);
if (node != -1 && !node_online(node))
node = -1;
/* Allocate per-root-bus (not per bus) arch-specific data.
* TODO: leak; this memory is never freed.
* It's arguable whether it's worth the trouble to care.
*/
sd = kzalloc(sizeof(*sd), GFP_KERNEL);
if (!sd) {
printk(KERN_WARNING "pci_bus %04x:%02x: "
"ignored (out of memory)\n", domain, busnum);
return NULL;
}
sd->domain = domain;
sd->node = node;
/*
* Maybe the desired pci bus has been already scanned. In such case
* it is unnecessary to scan the pci bus with the given domain,busnum.
*/
bus = pci_find_bus(domain, busnum);
if (bus) {
/*
* If the desired bus exits, the content of bus->sysdata will
* be replaced by sd.
*/
memcpy(bus->sysdata, sd, sizeof(*sd));
kfree(sd);
} else {
bus = pci_create_bus(NULL, busnum, &pci_root_ops, sd);
if (bus) {
get_current_resources(device, busnum, domain, bus);
bus->subordinate = pci_scan_child_bus(bus);
}
}
PCI: Set PCI-E Max Payload Size on fabric On a given PCI-E fabric, each device, bridge, and root port can have a different PCI-E maximum payload size. There is a sizable performance boost for having the largest possible maximum payload size on each PCI-E device. However, if improperly configured, fatal bus errors can occur. Thus, it is important to ensure that PCI-E payloads sends by a device are never larger than the MPS setting of all devices on the way to the destination. This can be achieved two ways: - A conservative approach is to use the smallest common denominator of the entire tree below a root complex for every device on that fabric. This means for example that having a 128 bytes MPS USB controller on one leg of a switch will dramatically reduce performances of a video card or 10GE adapter on another leg of that same switch. It also means that any hierarchy supporting hotplug slots (including expresscard or thunderbolt I suppose, dbl check that) will have to be entirely clamped to 128 bytes since we cannot predict what will be plugged into those slots, and we cannot change the MPS on a "live" system. - A more optimal way is possible, if it falls within a couple of constraints: * The top-level host bridge will never generate packets larger than the smallest TLP (or if it can be controlled independently from its MPS at least) * The device will never generate packets larger than MPS (which can be configured via MRRS) * No support of direct PCI-E <-> PCI-E transfers between devices without some additional code to specifically deal with that case Then we can use an approach that basically ignores downstream requests and focuses exclusively on upstream requests. In that case, all we need to care about is that a device MPS is no larger than its parent MPS, which allows us to keep all switches/bridges to the max MPS supported by their parent and eventually the PHB. In this case, your USB controller would no longer "starve" your 10GE Ethernet and your hotplug slots won't affect your global MPS. Additionally, the hotplugged devices themselves can be configured to a larger MPS up to the value configured in the hotplug bridge. To choose between the two available options, two PCI kernel boot args have been added to the PCI calls. "pcie_bus_safe" will provide the former behavior, while "pcie_bus_perf" will perform the latter behavior. By default, the latter behavior is used. NOTE: due to the location of the enablement, each arch will need to add calls to this function. This patch only enables x86. This patch includes a number of changes recommended by Benjamin Herrenschmidt. Tested-by: Jordan_Hargrave@dell.com Signed-off-by: Jon Mason <mason@myri.com> Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
2011-07-20 20:20:54 +00:00
/* After the PCI-E bus has been walked and all devices discovered,
* configure any settings of the fabric that might be necessary.
*/
if (bus) {
struct pci_bus *child;
list_for_each_entry(child, &bus->children, node) {
struct pci_dev *self = child->self;
if (!self)
continue;
pcie_bus_configure_settings(child, self->pcie_mpss);
}
PCI: Set PCI-E Max Payload Size on fabric On a given PCI-E fabric, each device, bridge, and root port can have a different PCI-E maximum payload size. There is a sizable performance boost for having the largest possible maximum payload size on each PCI-E device. However, if improperly configured, fatal bus errors can occur. Thus, it is important to ensure that PCI-E payloads sends by a device are never larger than the MPS setting of all devices on the way to the destination. This can be achieved two ways: - A conservative approach is to use the smallest common denominator of the entire tree below a root complex for every device on that fabric. This means for example that having a 128 bytes MPS USB controller on one leg of a switch will dramatically reduce performances of a video card or 10GE adapter on another leg of that same switch. It also means that any hierarchy supporting hotplug slots (including expresscard or thunderbolt I suppose, dbl check that) will have to be entirely clamped to 128 bytes since we cannot predict what will be plugged into those slots, and we cannot change the MPS on a "live" system. - A more optimal way is possible, if it falls within a couple of constraints: * The top-level host bridge will never generate packets larger than the smallest TLP (or if it can be controlled independently from its MPS at least) * The device will never generate packets larger than MPS (which can be configured via MRRS) * No support of direct PCI-E <-> PCI-E transfers between devices without some additional code to specifically deal with that case Then we can use an approach that basically ignores downstream requests and focuses exclusively on upstream requests. In that case, all we need to care about is that a device MPS is no larger than its parent MPS, which allows us to keep all switches/bridges to the max MPS supported by their parent and eventually the PHB. In this case, your USB controller would no longer "starve" your 10GE Ethernet and your hotplug slots won't affect your global MPS. Additionally, the hotplugged devices themselves can be configured to a larger MPS up to the value configured in the hotplug bridge. To choose between the two available options, two PCI kernel boot args have been added to the PCI calls. "pcie_bus_safe" will provide the former behavior, while "pcie_bus_perf" will perform the latter behavior. By default, the latter behavior is used. NOTE: due to the location of the enablement, each arch will need to add calls to this function. This patch only enables x86. This patch includes a number of changes recommended by Benjamin Herrenschmidt. Tested-by: Jordan_Hargrave@dell.com Signed-off-by: Jon Mason <mason@myri.com> Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
2011-07-20 20:20:54 +00:00
}
if (!bus)
kfree(sd);
if (bus && node != -1) {
#ifdef CONFIG_ACPI_NUMA
if (pxm >= 0)
dev_printk(KERN_DEBUG, &bus->dev,
"on NUMA node %d (pxm %d)\n", node, pxm);
#else
dev_printk(KERN_DEBUG, &bus->dev, "on NUMA node %d\n", node);
#endif
}
return bus;
}
int __init pci_acpi_init(void)
{
struct pci_dev *dev = NULL;
if (acpi_noirq)
return -ENODEV;
printk(KERN_INFO "PCI: Using ACPI for IRQ routing\n");
acpi_irq_penalty_init();
pcibios_enable_irq = acpi_pci_irq_enable;
pcibios_disable_irq = acpi_pci_irq_disable;
x86_init.pci.init_irq = x86_init_noop;
if (pci_routeirq) {
/*
* PCI IRQ routing is set up by pci_enable_device(), but we
* also do it here in case there are still broken drivers that
* don't use pci_enable_device().
*/
printk(KERN_INFO "PCI: Routing PCI interrupts for all devices because \"pci=routeirq\" specified\n");
for_each_pci_dev(dev)
acpi_pci_irq_enable(dev);
}
return 0;
}