* [PATCH pciutils 0/4] Support for PCI_FILL_PARENT
@ 2022-01-21 14:22 Pali Rohár
2022-01-21 14:22 ` [PATCH pciutils 1/4] libpci: Add new option PCI_FILL_PARENT Pali Rohár
` (4 more replies)
0 siblings, 5 replies; 8+ messages in thread
From: Pali Rohár @ 2022-01-21 14:22 UTC (permalink / raw)
To: Martin Mares, Bjorn Helgaas, Krzysztof Wilczyński,
Matthew Wilcox, linux-pci
Extend libpci API with a new option PCI_FILL_PARENT to fill parent
device for the current enumerated device. This can be useful in
situation when non-complaint PCI-to-PCI bridge-like device with Type 0
header is present in the system and behind this bridge are either
endpoint devices or another non-compliant bridges. This applies e.g.
for notoriously broken Galileo and Marvell PCI and PCIe devices.
lspci can will use PCI_FILL_PARENT information from the system if
config space does not provide enough information to build topology.
Pali Rohár (4):
libpci: Add new option PCI_FILL_PARENT
libpci: sysfs: Implement support for PCI_FILL_PARENT
lspci: Build tree based on PCI_FILL_PARENT information
lspci: Do not show -[00]- bus in tree output
lib/pci.h | 2 ++
lib/sysfs.c | 31 ++++++++++++++++++++++++
ls-tree.c | 69 +++++++++++++++++++++++++++++++++++++++++++++++++----
lspci.c | 2 +-
4 files changed, 98 insertions(+), 6 deletions(-)
--
2.20.1
^ permalink raw reply [flat|nested] 8+ messages in thread
* [PATCH pciutils 1/4] libpci: Add new option PCI_FILL_PARENT
2022-01-21 14:22 [PATCH pciutils 0/4] Support for PCI_FILL_PARENT Pali Rohár
@ 2022-01-21 14:22 ` Pali Rohár
2022-01-21 14:22 ` [PATCH pciutils 2/4] libpci: sysfs: Implement support for PCI_FILL_PARENT Pali Rohár
` (3 subsequent siblings)
4 siblings, 0 replies; 8+ messages in thread
From: Pali Rohár @ 2022-01-21 14:22 UTC (permalink / raw)
To: Martin Mares, Bjorn Helgaas, Krzysztof Wilczyński,
Matthew Wilcox, linux-pci
This change extends libpci and allows providers to fill parent pci_dev.
This is useful to retrieve topology as it is reported by the system itself.
---
lib/pci.h | 2 ++
1 file changed, 2 insertions(+)
diff --git a/lib/pci.h b/lib/pci.h
index c13387e2b4b1..de031f7a416b 100644
--- a/lib/pci.h
+++ b/lib/pci.h
@@ -148,6 +148,7 @@ struct pci_dev {
u8 prog_if; /* Programming interface for device_class */
u8 rev_id; /* Revision id */
u16 subsys_vendor_id, subsys_id; /* Subsystem vendor id and subsystem id */
+ struct pci_dev *parent; /* Parent device */
/* Fields used internally */
struct pci_access *access;
@@ -217,6 +218,7 @@ char *pci_get_string_property(struct pci_dev *d, u32 prop) PCI_ABI;
#define PCI_FILL_REVID 0x00040000
#define PCI_FILL_SUBSYS 0x00080000
#define PCI_FILL_DRIVER 0x00100000
+#define PCI_FILL_PARENT 0x00200000
void pci_setup_cache(struct pci_dev *, u8 *cache, int len) PCI_ABI;
--
2.20.1
^ permalink raw reply related [flat|nested] 8+ messages in thread
* [PATCH pciutils 2/4] libpci: sysfs: Implement support for PCI_FILL_PARENT
2022-01-21 14:22 [PATCH pciutils 0/4] Support for PCI_FILL_PARENT Pali Rohár
2022-01-21 14:22 ` [PATCH pciutils 1/4] libpci: Add new option PCI_FILL_PARENT Pali Rohár
@ 2022-01-21 14:22 ` Pali Rohár
2022-01-21 14:22 ` [PATCH pciutils 3/4] lspci: Build tree based on PCI_FILL_PARENT information Pali Rohár
` (2 subsequent siblings)
4 siblings, 0 replies; 8+ messages in thread
From: Pali Rohár @ 2022-01-21 14:22 UTC (permalink / raw)
To: Martin Mares, Bjorn Helgaas, Krzysztof Wilczyński,
Matthew Wilcox, linux-pci
---
lib/sysfs.c | 31 +++++++++++++++++++++++++++++++
1 file changed, 31 insertions(+)
diff --git a/lib/sysfs.c b/lib/sysfs.c
index 0dbd127ff53b..1db1b1a7752b 100644
--- a/lib/sysfs.c
+++ b/lib/sysfs.c
@@ -390,6 +390,37 @@ sysfs_fill_info(struct pci_dev *d, unsigned int flags)
{
done |= sysfs_get_resources(d);
}
+ if (flags & PCI_FILL_PARENT)
+ {
+ unsigned int domain, bus, dev, func;
+ char *path_abs, *path_canon, *name;
+ char path_rel[OBJNAMELEN];
+ struct pci_dev *parent;
+
+ /* Construct sysfs path for parent device */
+ sysfs_obj_name(d, "..", path_rel);
+ path_abs = realpath(path_rel, NULL);
+ name = path_abs ? strrchr(path_abs, '/') : NULL;
+ name = name ? name+1 : name;
+ parent = NULL;
+
+ if (name && sscanf(name, "%x:%x:%x.%d", &domain, &bus, &dev, &func) == 4 && domain <= 0x7fffffff)
+ for (parent = d->access->devices; parent; parent = parent->next)
+ if (parent->domain == (int)domain && parent->bus == bus && parent->dev == dev && parent->func == func)
+ break;
+
+ if (parent)
+ {
+ /* Check if parsed BDF address from parent sysfs device is really expected PCI device */
+ sysfs_obj_name(parent, ".", path_rel);
+ path_canon = realpath(path_rel, NULL);
+ if (path_canon && strcmp(path_canon, path_abs) == 0)
+ {
+ d->parent = parent;
+ done |= PCI_FILL_PARENT;
+ }
+ }
+ }
}
if (flags & PCI_FILL_PHYS_SLOT)
--
2.20.1
^ permalink raw reply related [flat|nested] 8+ messages in thread
* [PATCH pciutils 3/4] lspci: Build tree based on PCI_FILL_PARENT information
2022-01-21 14:22 [PATCH pciutils 0/4] Support for PCI_FILL_PARENT Pali Rohár
2022-01-21 14:22 ` [PATCH pciutils 1/4] libpci: Add new option PCI_FILL_PARENT Pali Rohár
2022-01-21 14:22 ` [PATCH pciutils 2/4] libpci: sysfs: Implement support for PCI_FILL_PARENT Pali Rohár
@ 2022-01-21 14:22 ` Pali Rohár
2022-01-21 14:22 ` [PATCH pciutils 4/4] lspci: Do not show -[00]- bus in tree output Pali Rohár
2022-01-21 14:51 ` [PATCH pciutils 0/4] Support for PCI_FILL_PARENT Martin Mareš
4 siblings, 0 replies; 8+ messages in thread
From: Pali Rohár @ 2022-01-21 14:22 UTC (permalink / raw)
To: Martin Mares, Bjorn Helgaas, Krzysztof Wilczyński,
Matthew Wilcox, linux-pci
Topology reported by system (libpci provider) may be different from
topology built based on primary/secondary/subordinate numbers from PCI
bridges by lspci.
This happens for example when some non-compliant PCI-to-PCI bridge
with Type 0 header (e.g. Marvell one) is available in the system.
So add additional edges reported by libpci when building tree in lspci.
---
ls-tree.c | 65 +++++++++++++++++++++++++++++++++++++++++++++++++++----
lspci.c | 2 +-
2 files changed, 62 insertions(+), 5 deletions(-)
diff --git a/ls-tree.c b/ls-tree.c
index aaa1ee99d9f2..fede581156e4 100644
--- a/ls-tree.c
+++ b/ls-tree.c
@@ -25,6 +25,19 @@ find_bus(struct bridge *b, unsigned int domain, unsigned int n)
return bus;
}
+static struct device *
+find_device(struct pci_dev *dd)
+{
+ struct device *d;
+
+ if (!dd)
+ return NULL;
+ for (d=first_dev; d; d=d->next)
+ if (d->dev == dd)
+ break;
+ return d;
+}
+
static struct bus *
new_bus(struct bridge *b, unsigned int domain, unsigned int n)
{
@@ -47,9 +60,20 @@ static void
insert_dev(struct device *d, struct bridge *b)
{
struct pci_dev *p = d->dev;
- struct bus *bus;
+ struct device *parent = NULL;
+ struct bus *bus = NULL;
+
+ if (p->known_fields & PCI_FILL_PARENT)
+ parent = find_device(p->parent);
- if (! (bus = find_bus(b, p->domain, p->bus)))
+ if (parent && parent->bridge)
+ {
+ bus = parent->bridge->first_bus;
+ if (!bus)
+ bus = new_bus(parent->bridge, p->domain, p->bus);
+ }
+
+ if (!bus && ! (bus = find_bus(b, p->domain, p->bus)))
{
struct bridge *c;
for (c=b->child; c; c=c->next)
@@ -113,14 +137,47 @@ grow_tree(void)
b->primary, b->secondary, b->subordinate);
}
}
+
+ /* Append additional bridges reported by libpci via d->parent */
+
+ for (d=first_dev; d; d=d->next)
+ {
+ struct device *parent = NULL;
+ if (d->dev->known_fields & PCI_FILL_PARENT)
+ parent = find_device(d->dev->parent);
+ if (!parent || parent->bridge)
+ continue;
+ b = xmalloc(sizeof(struct bridge));
+ b->domain = parent->dev->domain;
+ b->primary = parent->dev->bus;
+ b->secondary = d->dev->bus;
+ /* At this stage subordinate number is unknown, so set it to secondary bus number. */
+ b->subordinate = b->secondary;
+ *last_br = b;
+ last_br = &b->chain;
+ b->next = b->child = NULL;
+ b->first_bus = NULL;
+ b->last_bus = NULL;
+ b->br_dev = parent;
+ parent->bridge = b;
+ pacc->debug("Tree: bridge %04x:%02x:%02x.%d\n", b->domain,
+ parent->dev->bus, parent->dev->dev, parent->dev->func);
+ }
*last_br = NULL;
/* Create a bridge tree */
for (b=&host_bridge; b; b=b->chain)
{
- struct bridge *c, *best;
- best = NULL;
+ struct device *br_dev = b->br_dev;
+ struct bridge *c, *best = NULL;
+ struct device *parent = NULL;
+
+ if (br_dev && (br_dev->dev->known_fields & PCI_FILL_PARENT))
+ parent = find_device(br_dev->dev->parent);
+ if (parent)
+ best = parent->bridge;
+ if (!best)
for (c=&host_bridge; c; c=c->chain)
if (c != b && (c == &host_bridge || b->domain == c->domain) &&
b->primary >= c->secondary && b->primary <= c->subordinate &&
diff --git a/lspci.c b/lspci.c
index c4e6c93bc67a..b18ccba89049 100644
--- a/lspci.c
+++ b/lspci.c
@@ -143,7 +143,7 @@ scan_device(struct pci_dev *p)
d->config_cached += 64;
}
pci_setup_cache(p, d->config, d->config_cached);
- pci_fill_info(p, PCI_FILL_IDENT | PCI_FILL_CLASS | PCI_FILL_PROGIF | PCI_FILL_REVID | PCI_FILL_SUBSYS);
+ pci_fill_info(p, PCI_FILL_IDENT | PCI_FILL_CLASS | PCI_FILL_PROGIF | PCI_FILL_REVID | PCI_FILL_SUBSYS | (need_topology ? PCI_FILL_PARENT : 0));
return d;
}
--
2.20.1
^ permalink raw reply related [flat|nested] 8+ messages in thread
* [PATCH pciutils 4/4] lspci: Do not show -[00]- bus in tree output
2022-01-21 14:22 [PATCH pciutils 0/4] Support for PCI_FILL_PARENT Pali Rohár
` (2 preceding siblings ...)
2022-01-21 14:22 ` [PATCH pciutils 3/4] lspci: Build tree based on PCI_FILL_PARENT information Pali Rohár
@ 2022-01-21 14:22 ` Pali Rohár
2022-01-21 14:51 ` [PATCH pciutils 0/4] Support for PCI_FILL_PARENT Martin Mareš
4 siblings, 0 replies; 8+ messages in thread
From: Pali Rohár @ 2022-01-21 14:22 UTC (permalink / raw)
To: Martin Mares, Bjorn Helgaas, Krzysztof Wilczyński,
Matthew Wilcox, linux-pci
Secondary or subordinate bus cannot be zero. Zero value could indicate
either invalid secondary bus value or the fact that secondary bus value was
not filled or indicates non-compliant PCI-to-PCI bridge. This change makes
tree output better readable when bus numbers are not known or not provided.
---
ls-tree.c | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)
diff --git a/ls-tree.c b/ls-tree.c
index fede581156e4..f8154a2b034a 100644
--- a/ls-tree.c
+++ b/ls-tree.c
@@ -258,7 +258,9 @@ show_tree_dev(struct device *d, char *line, char *p)
for (b=&host_bridge; b; b=b->chain)
if (b->br_dev == d)
{
- if (b->secondary == b->subordinate)
+ if (b->secondary == 0)
+ p = tree_printf(line, p, "-");
+ else if (b->secondary == b->subordinate)
p = tree_printf(line, p, "-[%02x]-", b->secondary);
else
p = tree_printf(line, p, "-[%02x-%02x]-", b->secondary, b->subordinate);
--
2.20.1
^ permalink raw reply related [flat|nested] 8+ messages in thread
* Re: [PATCH pciutils 0/4] Support for PCI_FILL_PARENT
2022-01-21 14:22 [PATCH pciutils 0/4] Support for PCI_FILL_PARENT Pali Rohár
` (3 preceding siblings ...)
2022-01-21 14:22 ` [PATCH pciutils 4/4] lspci: Do not show -[00]- bus in tree output Pali Rohár
@ 2022-01-21 14:51 ` Martin Mareš
2022-01-21 15:21 ` Pali Rohár
4 siblings, 1 reply; 8+ messages in thread
From: Martin Mareš @ 2022-01-21 14:51 UTC (permalink / raw)
To: Pali Rohár
Cc: Bjorn Helgaas, Krzysztof Wilczyński, Matthew Wilcox, linux-pci
Hello!
> Extend libpci API with a new option PCI_FILL_PARENT to fill parent
> device for the current enumerated device. This can be useful in
> situation when non-complaint PCI-to-PCI bridge-like device with Type 0
> header is present in the system and behind this bridge are either
> endpoint devices or another non-compliant bridges. This applies e.g.
> for notoriously broken Galileo and Marvell PCI and PCIe devices.
> lspci can will use PCI_FILL_PARENT information from the system if
> config space does not provide enough information to build topology.
Looks reasonable, but please put a better explanation in pci.h
(in particular, mention that this is not guaranteed to be available).
When reading the implementation in the sysfs backend, I wonder how you
can guarantee that at the moment of parsing the child device, the parent
device is already known to libpci.
Martin
^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: [PATCH pciutils 0/4] Support for PCI_FILL_PARENT
2022-01-21 14:51 ` [PATCH pciutils 0/4] Support for PCI_FILL_PARENT Martin Mareš
@ 2022-01-21 15:21 ` Pali Rohár
2022-01-21 15:28 ` Martin Mareš
0 siblings, 1 reply; 8+ messages in thread
From: Pali Rohár @ 2022-01-21 15:21 UTC (permalink / raw)
To: Martin Mareš
Cc: Bjorn Helgaas, Krzysztof Wilczyński, Matthew Wilcox, linux-pci
On Friday 21 January 2022 15:51:34 Martin Mareš wrote:
> Hello!
>
> > Extend libpci API with a new option PCI_FILL_PARENT to fill parent
> > device for the current enumerated device. This can be useful in
> > situation when non-complaint PCI-to-PCI bridge-like device with Type 0
> > header is present in the system and behind this bridge are either
> > endpoint devices or another non-compliant bridges. This applies e.g.
> > for notoriously broken Galileo and Marvell PCI and PCIe devices.
> > lspci can will use PCI_FILL_PARENT information from the system if
> > config space does not provide enough information to build topology.
>
> Looks reasonable, but please put a better explanation in pci.h
> (in particular, mention that this is not guaranteed to be available).
Ok!
> When reading the implementation in the sysfs backend, I wonder how you
> can guarantee that at the moment of parsing the child device, the parent
> device is already known to libpci.
All devices are parsed in sysfs_scan() function. And additional
information, including this "parent" is filled in sysfs_fill_info()
function. sysfs_scan() itself does not call sysfs_fill_info().
So prior calling pci_fill_info() it is required to call pci_scan_bus()
to have ->parent member filled.
^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: [PATCH pciutils 0/4] Support for PCI_FILL_PARENT
2022-01-21 15:21 ` Pali Rohár
@ 2022-01-21 15:28 ` Martin Mareš
0 siblings, 0 replies; 8+ messages in thread
From: Martin Mareš @ 2022-01-21 15:28 UTC (permalink / raw)
To: Pali Rohár
Cc: Bjorn Helgaas, Krzysztof Wilczyński, Matthew Wilcox, linux-pci
> All devices are parsed in sysfs_scan() function. And additional
> information, including this "parent" is filled in sysfs_fill_info()
> function. sysfs_scan() itself does not call sysfs_fill_info().
>
> So prior calling pci_fill_info() it is required to call pci_scan_bus()
> to have ->parent member filled.
Ah, OK. My fault.
It's OK then.
Have a nice fortnight
--
Martin `MJ' Mareš <mj@ucw.cz> http://mj.ucw.cz/
United Computer Wizards, Prague, Czech Republic, Europe, Earth, Universe
A LISP programmer knows value of everything, but cost of nothing.
^ permalink raw reply [flat|nested] 8+ messages in thread
end of thread, other threads:[~2022-01-21 15:29 UTC | newest]
Thread overview: 8+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2022-01-21 14:22 [PATCH pciutils 0/4] Support for PCI_FILL_PARENT Pali Rohár
2022-01-21 14:22 ` [PATCH pciutils 1/4] libpci: Add new option PCI_FILL_PARENT Pali Rohár
2022-01-21 14:22 ` [PATCH pciutils 2/4] libpci: sysfs: Implement support for PCI_FILL_PARENT Pali Rohár
2022-01-21 14:22 ` [PATCH pciutils 3/4] lspci: Build tree based on PCI_FILL_PARENT information Pali Rohár
2022-01-21 14:22 ` [PATCH pciutils 4/4] lspci: Do not show -[00]- bus in tree output Pali Rohár
2022-01-21 14:51 ` [PATCH pciutils 0/4] Support for PCI_FILL_PARENT Martin Mareš
2022-01-21 15:21 ` Pali Rohár
2022-01-21 15:28 ` Martin Mareš
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).