]> mj.ucw.cz Git - pciutils.git/log
pciutils.git
8 years agolspci: Decode Precision Time Measurement capabiltity
Yong, Jonathan [Tue, 10 May 2016 03:15:55 +0000 (03:15 +0000)]
lspci: Decode Precision Time Measurement capabiltity

Section 7.32 Precision Time Management (or Measurement) from the
PCI Express Base 3.1 specification is an optional Extended Capability
for discovering and controlling the distribution of a PTM Hierarchy.

Signed-off-by: Yong, Jonathan <jonathan.yong@intel.com>
8 years agoReleased as 3.4.1. v3.4.1
Martin Mares [Sun, 3 Jan 2016 14:19:41 +0000 (15:19 +0100)]
Released as 3.4.1.

8 years agopci.ids updated to today's snapshot
Martin Mares [Sun, 3 Jan 2016 14:18:55 +0000 (15:18 +0100)]
pci.ids updated to today's snapshot

8 years agoAdd lspci support for Enhanced Allocation Capability.
David Daney [Wed, 23 Dec 2015 00:33:51 +0000 (16:33 -0800)]
Add lspci support for Enhanced Allocation Capability.

The PCISIG recently added the Enhanced Allocation Capability.  Decode
it in lspci.

8 years agolspci: Decode DevCap SlotPowerLimit for all components with Upstream Ports
Bjorn Helgaas [Thu, 10 Dec 2015 19:50:01 +0000 (13:50 -0600)]
lspci: Decode DevCap SlotPowerLimit for all components with Upstream Ports

The SlotPowerLimit in the Slot Capability indicates how much power the slot
can supply to a downstream device.  A Root Port or Switch Downstream Port
communicates the limit via a Set_Slot_Power_Limit Message on the link.  The
component on the other end of the link copies the limit from the message to
the Captured Slot Power Limit in its Device Capability [see PCIe r3.0, sec
2.2.8.5].

The Captured SlotPowerLimit is relevant for all devices on the downstream
end of a Link.  This includes Endpoints and Bridges as well as
Switch Upstream Ports.

Decode the DevCap Captured SlotPowerLimit for Endpoints and Bridges as well
as Switch Upstream Ports.

Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
9 years agoOmit unnecessary commas in PASID/PRI capabilities
Martin Mares [Tue, 27 Oct 2015 11:19:10 +0000 (12:19 +0100)]
Omit unnecessary commas in PASID/PRI capabilities

9 years agoDecode PASID and PRI extended capabilities
David Woodhouse [Thu, 15 Oct 2015 18:44:16 +0000 (19:44 +0100)]
Decode PASID and PRI extended capabilities

9 years agoREADME: SourceForge mirrors replaced by GitHub
Martin Mares [Sat, 19 Sep 2015 13:50:26 +0000 (15:50 +0200)]
README: SourceForge mirrors replaced by GitHub

9 years agoMaint: Remember to push to Github, too
Martin Mares [Sat, 19 Sep 2015 13:18:06 +0000 (15:18 +0200)]
Maint: Remember to push to Github, too

9 years agoReleased as 3.4.0. v3.4.0
Martin Mares [Mon, 14 Sep 2015 15:53:11 +0000 (17:53 +0200)]
Released as 3.4.0.

9 years agoUpdated pci.ids to today's snapshot
Martin Mares [Mon, 14 Sep 2015 15:49:29 +0000 (17:49 +0200)]
Updated pci.ids to today's snapshot

9 years agolib/configure: Also accept host tuples without vendor string
Felix Janda [Sun, 12 Apr 2015 07:16:28 +0000 (09:16 +0200)]
lib/configure: Also accept host tuples without vendor string

Based on patch from
https://bugs.gentoo.org/show_bug.cgi?id=425022

9 years agolspci calls pci_fill_info() only as needed
Martin Mares [Mon, 14 Sep 2015 15:44:45 +0000 (17:44 +0200)]
lspci calls pci_fill_info() only as needed

Previously, lspci always asked for all attributes, even in terse
mode where most of them are not shown.

9 years agoSysfs: Read failures of optional attributes are not fatal
Martin Mares [Mon, 14 Sep 2015 15:43:04 +0000 (17:43 +0200)]
Sysfs: Read failures of optional attributes are not fatal

Ameya Palande reported that with some kernels, reads of such
attributes fail on some hardware. He suggested to ignore read
failures completely, but I decided to turn the errors into
warnings in such cases. At least, the user will know that something
fishy is going on.

9 years agoOops, numa_node should not be a mandatory attribute
Martin Mares [Mon, 14 Sep 2015 15:42:11 +0000 (17:42 +0200)]
Oops, numa_node should not be a mandatory attribute

9 years agoUpdated fixed-version references to pci_fill_info
Martin Mares [Mon, 14 Sep 2015 15:24:29 +0000 (17:24 +0200)]
Updated fixed-version references to pci_fill_info

9 years agoNUMA node scanning is now done in an ABI-compatible way
Martin Mares [Mon, 14 Sep 2015 15:00:28 +0000 (17:00 +0200)]
NUMA node scanning is now done in an ABI-compatible way

The numa_node field was moved to the end of the public part of
struct pci_dev. As usually, it has to be requested using the
PCI_FILL_NUMA_NODE and pci_fill_info() is versioned.

9 years agoWhitespace fixes
Martin Mares [Mon, 14 Sep 2015 14:48:32 +0000 (16:48 +0200)]
Whitespace fixes

9 years agoImproved reporting of NUMA nodes
Martin Mares [Mon, 14 Sep 2015 14:46:49 +0000 (16:46 +0200)]
Improved reporting of NUMA nodes

They are also printed in "-vv" and "-nv" modes now.

9 years agoReport NUMA node in lspci -v
Matthew Wilcox [Wed, 22 Apr 2015 20:27:21 +0000 (16:27 -0400)]
Report NUMA node in lspci -v

In multi-socket systems, it's useful to see which node a particular
PCI device belongs to.  Linux provides this information through sysfs,
but some users don't like poking through sysfs themselves to find it,
and it's pretty straightforward to report it in lspci.

I should note that when there is no NUMA node for a particular device,
Linux reports -1.  I've chosen to continue that convention in pciutils,
and simply omit the information if the device does not belong to a NUMA
node (eg on single-socket systems, or devices which are not preferentially
attached to a particular node, like Nehalem-based systems).

9 years agoFixed naming inconsistency in man page: slot vs. device
Martin Mares [Mon, 14 Sep 2015 14:40:42 +0000 (16:40 +0200)]
Fixed naming inconsistency in man page: slot vs. device

Thanks to Robert Urban for pointing it out.

9 years agoFix broken backward compat struct translation for pci filters
Lucas Stach [Mon, 13 Apr 2015 18:42:25 +0000 (20:42 +0200)]
Fix broken backward compat struct translation for pci filters

This seems to be a copy&paste error in both directions of the
compat translation.

Signed-off-by: Lucas Stach <dev@lynxeye.de>
9 years agoFixing up release 3.3.1 to contain new pci.ids v3.3.1
Martin Mares [Thu, 9 Apr 2015 14:42:40 +0000 (16:42 +0200)]
Fixing up release 3.3.1 to contain new pci.ids

9 years agoUpdate pci.ids to today's snapshot
Martin Mares [Thu, 9 Apr 2015 14:42:12 +0000 (16:42 +0200)]
Update pci.ids to today's snapshot

9 years agoReleased as 3.3.1.
Martin Mares [Thu, 9 Apr 2015 14:36:55 +0000 (16:36 +0200)]
Released as 3.3.1.

9 years agopread.h: Remove support for libc5
Felix Janda [Sun, 29 Mar 2015 08:59:32 +0000 (09:59 +0100)]
pread.h: Remove support for libc5

The support code for libc5 breaks building on linux i386 with
other libcs that don't define __GLIBC__.

9 years agoMaint: Make sure that there are no Vim swap files in the tarball
Martin Mares [Thu, 22 Jan 2015 21:47:16 +0000 (22:47 +0100)]
Maint: Make sure that there are no Vim swap files in the tarball

9 years agoFixed memory allocation bug in name cache path code
Martin Mares [Thu, 22 Jan 2015 21:44:49 +0000 (22:44 +0100)]
Fixed memory allocation bug in name cache path code

9 years agoImproved listing of vendor-specific information
Martin Mares [Thu, 22 Jan 2015 21:42:33 +0000 (22:42 +0100)]
Improved listing of vendor-specific information

9 years agoAdded test case for virtio
Martin Mares [Thu, 22 Jan 2015 21:31:24 +0000 (22:31 +0100)]
Added test case for virtio

9 years agoAdd virtio vendor capability support
Gerd Hoffmann [Wed, 21 Jan 2015 15:35:04 +0000 (16:35 +0100)]
Add virtio vendor capability support

virtio uses vendor-specific capabilities to specify the location of
the virtio register ranges.  The specification can be found here:

http://docs.oasis-open.org/virtio/virtio/v1.0/cs01/virtio-v1.0-cs01.html#x1-690004

This patch adds support for decoding these capabilities to lspci.

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
9 years agoFix small memory leak in names-cache
Martin Mares [Fri, 12 Dec 2014 17:18:35 +0000 (18:18 +0100)]
Fix small memory leak in names-cache

Signed-off-by: Tomáš Chvátal <tchvatal@suse.cz>
---
 lib/names-cache.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/lib/names-cache.c b/lib/names-cache.c
index 90a6454..c97ea30 100644
--- a/lib/names-cache.c
+++ b/lib/names-cache.c
@@ -39,7 +39,8 @@ static char *get_cache_name(struct pci_access *a)
   buf = pci_malloc(a, strlen(pw->pw_dir) + strlen(name+1) + 1);
   sprintf(buf, "%s%s", pw->pw_dir, name+1);
   pci_set_param_internal(a, "net.cache_name", buf, 0);
-  return buf;
+  pci_mfree(buf);
+  return pci_get_param(a, "net.cache_name");
 }

 int
--
2.1.3

9 years agoMaintenance scripts now upload to kernel.org automatically
Martin Mares [Mon, 10 Nov 2014 14:01:01 +0000 (15:01 +0100)]
Maintenance scripts now upload to kernel.org automatically

9 years agoChangeLog: Fix formatting so that it is parseable by maint/release v3.3.0
Martin Mares [Mon, 10 Nov 2014 13:46:15 +0000 (14:46 +0100)]
ChangeLog: Fix formatting so that it is parseable by maint/release

9 years agoUpdated PCI_LIB_VERSION
Martin Mares [Mon, 10 Nov 2014 13:40:50 +0000 (14:40 +0100)]
Updated PCI_LIB_VERSION

9 years agoReleased as 3.3.0.
Martin Mares [Mon, 10 Nov 2014 13:38:10 +0000 (14:38 +0100)]
Released as 3.3.0.

9 years agoUpdated pci.ids to today's snapshot.
Martin Mares [Mon, 10 Nov 2014 13:34:12 +0000 (14:34 +0100)]
Updated pci.ids to today's snapshot.

10 years agoChangeLog: Mention filtering by class
Martin Mares [Sun, 2 Nov 2014 11:14:56 +0000 (12:14 +0100)]
ChangeLog: Mention filtering by class

10 years agoBackward ABI compatibility for new filters and pci_fill_info
Martin Mares [Sun, 2 Nov 2014 11:11:05 +0000 (12:11 +0100)]
Backward ABI compatibility for new filters and pci_fill_info

This is tricky, because we have to translate between old and new
format of struct pci_filter. At least, I added several RFU fields
so this hopefully won't have to happen again soon.

10 years agolspci: Add ability to filter by class code
Matthew Wilcox [Tue, 30 Sep 2014 18:02:52 +0000 (14:02 -0400)]
lspci: Add ability to filter by class code

Extend the 'filter by device ID' functionality to allow optional
specification of a device ID.  For example, to list all USB controllers
in the system made by Intel, specify:

lspci -d 8086::0c03

Signed-off-by: Matthew Wilcox <matthew.r.wilcox@intel.com>
10 years agoUpdated ChangeLog
Martin Mares [Sun, 2 Nov 2014 10:35:29 +0000 (11:35 +0100)]
Updated ChangeLog

10 years agoClean up man pages
Martin Mares [Sun, 2 Nov 2014 10:23:20 +0000 (11:23 +0100)]
Clean up man pages

We do not call .IX as it's not defined in mandoc.
Ellipsis at the end of line is protected by \&.

Inspired by Debian patches for the following bugs:
https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=674708
https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=675087

10 years agoConfigure: HWDB is specific for Linux
Martin Mares [Sun, 2 Nov 2014 10:10:45 +0000 (11:10 +0100)]
Configure: HWDB is specific for Linux

10 years agoChangeLog: Fixed a typo
Martin Mares [Sat, 1 Nov 2014 18:12:26 +0000 (19:12 +0100)]
ChangeLog: Fixed a typo

10 years agoUpdated the READMEs
Martin Mares [Sat, 1 Nov 2014 18:11:11 +0000 (19:11 +0100)]
Updated the READMEs

10 years agoRewritten support for UDEV's HWDB
Martin Mares [Sat, 1 Nov 2014 17:38:08 +0000 (18:38 +0100)]
Rewritten support for UDEV's HWDB

HWDB is now handled in a way very similar to the DNS resolver.

The interface lives in a separate source file (lib/names-hwdb.c),
results of lookups are cached. Use of HWDB can be disabled either
by passing PCI_LOOKUP_NO_HWDB or by setting the hwdb.disabled
configuration parameter.

Also, there should be no more leaks of libudev's structures.

10 years agolibpci: pci_id_lookup - add udev/hwdb support
Tom Gundersen [Mon, 30 Dec 2013 18:59:53 +0000 (19:59 +0100)]
libpci: pci_id_lookup - add udev/hwdb support

This lets you select hwdb support at compile time.

hwdb is an efficient hardware database shipped with recent versions of udev. It contains
among other sources pci.ids so querying hwdb rather than reading pci.ids directly should give
the same result.

Ideally Linux distros using udev could stop shipping pci.ids, but use hwdb as the only source
of this information, which this patch allows.

Cc: Martin Mares <mj@ucw.cz>
10 years agoUpdated pci.ids to today's snapshot
Martin Mares [Sat, 1 Nov 2014 16:33:05 +0000 (17:33 +0100)]
Updated pci.ids to today's snapshot

10 years agoWork around problems with symbol versioning in globally optimizing GCC
Martin Mares [Sat, 1 Nov 2014 16:28:28 +0000 (17:28 +0100)]
Work around problems with symbol versioning in globally optimizing GCC

10 years agoDevice labels must be requested via PCI_FILL_LABELS
Martin Mares [Sat, 1 Nov 2014 16:11:17 +0000 (17:11 +0100)]
Device labels must be requested via PCI_FILL_LABELS

The previous implementation handled labels differently from all
other device properties, which was illogical.

10 years agoManual: Mention Darwin in the list of access methods
Martin Mares [Sat, 1 Nov 2014 15:59:19 +0000 (16:59 +0100)]
Manual: Mention Darwin in the list of access methods

10 years agoDarwin: Yet more renaming
Martin Mares [Sat, 1 Nov 2014 15:56:51 +0000 (16:56 +0100)]
Darwin: Yet more renaming

10 years agoDarwin: More renames
Martin Mares [Sat, 1 Nov 2014 15:56:22 +0000 (16:56 +0100)]
Darwin: More renames

10 years agoDarwin: Renamed to darwin.c
Martin Mares [Sat, 1 Nov 2014 15:55:17 +0000 (16:55 +0100)]
Darwin: Renamed to darwin.c

10 years agoDarwin: Re-indent to match the rest of libpci and update file header
Martin Mares [Sat, 1 Nov 2014 15:54:34 +0000 (16:54 +0100)]
Darwin: Re-indent to match the rest of libpci and update file header

I hope that I got the attribution right.

10 years agoBetter work-around for missing symbol versioning on Darwin
Martin Mares [Sat, 1 Nov 2014 15:49:42 +0000 (16:49 +0100)]
Better work-around for missing symbol versioning on Darwin

10 years agoFix building shared libraries on Darwin
Richard Yao [Mon, 20 Oct 2014 03:13:06 +0000 (23:13 -0400)]
Fix building shared libraries on Darwin

The original patch by Apple did not support building shared libraries on
Darwin. This corrects that oversight. It also fixes a few other
miscellaneous issues like incorrect platform detection and the lack of
an entry in the README file.

Signed-off-by: Richard Yao <ryao@gentoo.org>
10 years agoFix whitespace errors
Martin Mares [Sat, 1 Nov 2014 15:39:32 +0000 (16:39 +0100)]
Fix whitespace errors

10 years agoImport initial Darwin Support from Apple
Richard Yao [Mon, 20 Oct 2014 03:13:05 +0000 (23:13 -0400)]
Import initial Darwin Support from Apple

Apple published a patch that adds Darwin support to pciutils. It is not
complete, but it does not appear to cause a build failure on other
platforms. For the sake of attribution purposes, it is being kept
separate from additional changes to make pciutils build properly on
Darwin.

https://www.opensource.apple.com/source/IOPCIFamily/IOPCIFamily-224.92.1/tools/pciutils3.2.0.patch.c

This patch differs slightly from Apple's original patch by omitting the
Makefile.rej that Apple had incorrectly included in the patch that it
placed on its server.

Signed-off-by: Richard Yao <ryao@gentoo.org>
10 years agoFix endianity detection on Solaris
Martin Mares [Sat, 1 Nov 2014 15:31:49 +0000 (16:31 +0100)]
Fix endianity detection on Solaris

Patch by Philip Brown.

10 years agolspci: Correct Root Capabilities "CRS Software Visibility" bit
Bjorn Helgaas [Tue, 2 Sep 2014 04:03:46 +0000 (22:03 -0600)]
lspci: Correct Root Capabilities "CRS Software Visibility" bit

lspci incorrectly tests bit 4, not bit 0, for "CRS Software Visibility" in
the Root Capabilities register, so it shows "RootCap: CRSVisible-" even for
devices that do support Software Visibility.

Use the correct definition for PCI_EXP_RTCAP_CRSVIS.

Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
10 years agolspci: Decode Interrupt Pin and Interrupt Line for PCI-PCI bridges
Bjorn Helgaas [Sat, 8 Mar 2014 00:26:17 +0000 (17:26 -0700)]
lspci: Decode Interrupt Pin and Interrupt Line for PCI-PCI bridges

Bridges can implement interrupt pins, so decode this information.  See
PCI-to-PCI Bridge spec r1.2, sec 3.2.5.17.

Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
10 years agoClean up reading of device labels
Martin Mares [Sat, 1 Nov 2014 15:21:49 +0000 (16:21 +0100)]
Clean up reading of device labels

10 years agolspci: Show device name as exported by BIOS (sysfs label attribute) (V2)
Thomas Renninger [Tue, 28 Jan 2014 07:51:02 +0000 (08:51 +0100)]
lspci: Show device name as exported by BIOS (sysfs label attribute) (V2)

The Device name of a PCI or PCI Express device under OS may be exported via
ACPI _DSM function with function index 7.
This allows to connect a described PCI device in the platform documentation
or as labeled on the chassis with PCI devices shown via lspci.
The kernel already exports this string through sysfs under a PCI device through
the "label" sysfs attribute.

This patch reads the device name if available and shows it to the user.
Real world examples:
  Device Name: "USB HS EHCI Controller #2 #3"
  Device Name: "USB HS EHCI Controller #1"
  Device Name: "SATA Controller #1"
  Device Name: "Onboard LAN #1"
  Device Name: "Onboard LAN #2"
  Device Name: "Onboard Video (PILOT-3)"

Compare with PCI Firmware Spec v3.1 chapter 4.6.7 and
ACPI spec v5.0 chapter 9.14.1

The DeviceName is not shown by default, but starting from first verbose
parameter (-v).

V2: - Free label string if allocated
    - Enhance changelog

Signed-off-by: Thomas Renninger <trenn@suse.de>
CC: linux-pci@vger.kernel.org
10 years agoAvoid C99 constructions
François Revol [Wed, 18 Dec 2013 13:11:42 +0000 (14:11 +0100)]
Avoid C99 constructions

This fixes building pciutils on Haiku.

Signed-off-by: François Revol <revol@free.fr>
10 years ago<asm/io.h> does not exist any more
Martin Mares [Sat, 1 Nov 2014 15:17:58 +0000 (16:17 +0100)]
<asm/io.h> does not exist any more

Thanks to John Spencer for bringing this to attention.

10 years agoDo not forget to include <stdio.h>
Martin Mares [Sat, 1 Nov 2014 15:17:19 +0000 (16:17 +0100)]
Do not forget to include <stdio.h>

Suggested by John Spencer.

10 years agolspci: Decode ASPMOptComp bit in LnkCap register
Martin Mares [Tue, 25 Feb 2014 10:37:28 +0000 (11:37 +0100)]
lspci: Decode ASPMOptComp bit in LnkCap register

Patch by Robert Elliott from HP.

10 years agoFixed a typo in messages printed in map mode
Martin Mares [Tue, 10 Dec 2013 19:54:34 +0000 (20:54 +0100)]
Fixed a typo in messages printed in map mode

10 years agomaint/tag-release: Use my kernel.org key to sign releases v3.2.1
Martin Mares [Sun, 10 Nov 2013 12:40:07 +0000 (13:40 +0100)]
maint/tag-release: Use my kernel.org key to sign releases

10 years agoReleased as 3.2.1.
Martin Mares [Sun, 10 Nov 2013 12:22:17 +0000 (13:22 +0100)]
Released as 3.2.1.

10 years agoUpdated pci.ids to today's snapshot
Martin Mares [Sun, 10 Nov 2013 12:17:26 +0000 (13:17 +0100)]
Updated pci.ids to today's snapshot

10 years agosysfs: Do not warn on incomplete slot addresses
Martin Mares [Sun, 10 Nov 2013 11:39:19 +0000 (12:39 +0100)]
sysfs: Do not warn on incomplete slot addresses

Based on a patch by Zheng Huai Cheng <zhenghch@linux.vnet.ibm.com>.

10 years agolspci: Drop PCIe LnkCtl "Retrain Link" decoding
Bjorn Helgaas [Mon, 9 Sep 2013 23:14:21 +0000 (17:14 -0600)]
lspci: Drop PCIe LnkCtl "Retrain Link" decoding

Per PCIe spec r3.0, Table 7-16, the Retrain Link bit is writable but
always returns 0 when read, so decoding it gives no useful information.

Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
10 years agolspci: Decode PCIe LnkCtl "Read Completion Boundary" for PCIe-to-PCI bridges
Bjorn Helgaas [Mon, 9 Sep 2013 23:14:14 +0000 (17:14 -0600)]
lspci: Decode PCIe LnkCtl "Read Completion Boundary" for PCIe-to-PCI bridges

The PCIe spec (r3.0, Table 7-16) says the Read Completion Boundary is valid
for Root Ports, Endpoints, and Bridges.  I only added decoding for PCIe-to-
PCI/PCI-X bridges because the RCB of a Bridge indicates the RCB of the
upstream Root Port, so I don't think it makes sense for PCI-to-PCIe
bridges.

Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
10 years agolspci: Clarify "PCIe-to-PCI/PCI-X" desc and Bridge Retry Config Enable
Bjorn Helgaas [Mon, 9 Sep 2013 23:14:07 +0000 (17:14 -0600)]
lspci: Clarify "PCIe-to-PCI/PCI-X" desc and Bridge Retry Config Enable

The PCI_EXP_TYPE_PCI_BRIDGE type is a PCIe to PCI/PCI-X bridge, so be a bit
more complete in the comment and printed device type.  Also, per PCIe spec
r3.0, Table 7-14, the PCIe Device Control "Bridge Configuration Retry
Enable" bit only applies to PCIe-to-PCI/PCI-X bridges; it does not apply to
PCI-to-PCIe bridges.

Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
10 years agolspci: Decode PCIe DevCap/DevCtl FLReset only for Endpoints
Bjorn Helgaas [Mon, 9 Sep 2013 23:14:00 +0000 (17:14 -0600)]
lspci: Decode PCIe DevCap/DevCtl FLReset only for Endpoints

The PCIe Device Capabilities and Control bits related to Function
Level Reset are valid only for Endpoints, so only decode them in
that case.

Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
10 years agolspci: Decode PCIe DevCap "Acceptable Latencies" only for Endpoints
Bjorn Helgaas [Mon, 9 Sep 2013 23:13:53 +0000 (17:13 -0600)]
lspci: Decode PCIe DevCap "Acceptable Latencies" only for Endpoints

The PCIe Device Capabilities "Endpoint L0s Acceptable Latency" and
"Endpoint L1 Acceptable Latency" are defined only for Endpoint functions,
so don't display them unless this is an endpoint.

Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
10 years agolspci: Label PCIe LnkCap "L0s Exit Latency" as "L0s", not "L0"
Bjorn Helgaas [Mon, 9 Sep 2013 23:13:45 +0000 (17:13 -0600)]
lspci: Label PCIe LnkCap "L0s Exit Latency" as "L0s", not "L0"

The PCIe Link Capabilities "L0s Exit Latency" is the latency to exit
L0s, not L0, so label it "L0s" instead of "L0".  This matches the
way we label the Device Capabilities "Endpoint L0s Acceptable Latency"
field as "Latency L0s".  This also adds "Exit" to the description to
help distinguish it from the "Acceptable Latency" fields in the
Device Capabilities register.

Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
10 years agolspci: Decode PCIe Link registers only for devices with links
Bjorn Helgaas [Wed, 28 Aug 2013 19:08:34 +0000 (13:08 -0600)]
lspci: Decode PCIe Link registers only for devices with links

Root Complex Integrated Endpoints and Root Complex Event Collectors do not
have links and are not permitted to implement Link or Link 2 registers,
per PCIe spec r3.0, sec 1.3.2.3.  Decoding them is useless and misleading.

Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
11 years agolspci: Make output of L1 PM cap consistent with other caps
Martin Mares [Tue, 11 Jun 2013 07:49:23 +0000 (09:49 +0200)]
lspci: Make output of L1 PM cap consistent with other caps

11 years agoAdded a test case for the L1 PM capability
Martin Mares [Tue, 11 Jun 2013 07:47:58 +0000 (09:47 +0200)]
Added a test case for the L1 PM capability

Based on data contributed by David Box.

11 years agolspci: Add L1 PM Substate capability reporting
David Box [Tue, 11 Jun 2013 01:09:59 +0000 (18:09 -0700)]
lspci: Add L1 PM Substate capability reporting

Expose available L1 substate capabilities that can enable lower power
consumption when a PCIe Link is idle.

Signed-off-by: David Box <david.e.box@linux.intel.com>
11 years agolspci: Fully decode ASPM support from Link Capabilities
Bjorn Helgaas [Fri, 17 May 2013 19:48:19 +0000 (13:48 -0600)]
lspci: Fully decode ASPM support from Link Capabilities

The ASPM Support field in Link Capabilities is two bits, and all four
possible encodings are defined as of PCIe spec r3.0.  Previously, lspci
only decoded values 1, 2, and 3.  This adds 0, so lspci will show "ASPM
not supported" instead of "ASPM unknown".

Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
11 years agolspci: Display CardBus bridge capabilities
Matthew Wilcox [Sat, 20 Apr 2013 13:45:32 +0000 (07:45 -0600)]
lspci: Display CardBus bridge capabilities

CardBus bridges can have capabilities, but the CAP_PTR register is at a
different location.  This one has Power Management:

CardBus bridge: O2 Micro, Inc. OZ711SP1 Memory CardBus Controller (rev 01)
+       Capabilities: [a0] Power Management version 2

11 years agomaint/KORG-NOTES: Private maintainer's notes about kernel.org
Martin Mares [Fri, 19 Apr 2013 13:25:06 +0000 (15:25 +0200)]
maint/KORG-NOTES: Private maintainer's notes about kernel.org

11 years agomaint/release: metalab.unc.edu does not accept submissions any longer
Martin Mares [Fri, 19 Apr 2013 13:21:48 +0000 (15:21 +0200)]
maint/release: metalab.unc.edu does not accept submissions any longer

11 years agoReleased as 3.2.0. v3.2.0
Martin Mares [Fri, 19 Apr 2013 12:59:12 +0000 (14:59 +0200)]
Released as 3.2.0.

11 years agoUpdated pci.ids to today's snapshot
Martin Mares [Fri, 19 Apr 2013 12:47:32 +0000 (14:47 +0200)]
Updated pci.ids to today's snapshot

11 years agolspci: Better filtering of duplicate kernel module names
Martin Mares [Fri, 19 Apr 2013 11:33:52 +0000 (13:33 +0200)]
lspci: Better filtering of duplicate kernel module names

It was implemented only for reading modules.pcimap, but it turned out
that it is necessary for libkmod, too, so we have switched to a common
implementation.

11 years agoMinor cleanup of the previous patch
Martin Mares [Mon, 1 Apr 2013 20:11:04 +0000 (22:11 +0200)]
Minor cleanup of the previous patch

11 years agoinstall: split out pcilib install to a dedicated target
Kevin Pyle [Tue, 31 Jan 2012 15:50:43 +0000 (10:50 -0500)]
install: split out pcilib install to a dedicated target

The "install" target manages one set of files, and the "install-pcilib"
target manages a different set.  They both install the pci library though
so if you try to run `make -j install install-pcilib`, things randomly
fail.  So split out the commonly installed files into a dedicated target.

11 years agolib/configure: match all freebsd os's
Marek Cruz [Tue, 31 Jan 2012 15:50:47 +0000 (10:50 -0500)]
lib/configure: match all freebsd os's

When using targets like "i686-pc-freebsd7.1", the configure script fails
to match for the freebsd target because it only expects "freebsd".  Add
a glob to match all freebsd targets.

11 years agoSquash an unused variable warning
Martin Mares [Mon, 1 Apr 2013 19:25:51 +0000 (21:25 +0200)]
Squash an unused variable warning

11 years agosetpci: ~0UL should be ~0U, no longs are involved anymore
Martin Mares [Mon, 1 Apr 2013 19:24:03 +0000 (21:24 +0200)]
setpci: ~0UL should be ~0U, no longs are involved anymore

11 years agoConfigure: pkg-config can be overridden by $PKG_CONFIG
Martin Mares [Mon, 1 Apr 2013 19:23:44 +0000 (21:23 +0200)]
Configure: pkg-config can be overridden by $PKG_CONFIG

11 years agoOn newer Linux systems, use libkmod to look up kernel modules
Martin Mares [Mon, 1 Apr 2013 13:44:12 +0000 (15:44 +0200)]
On newer Linux systems, use libkmod to look up kernel modules

With modutils built upon libkmod, modules.pcimap does not exist
any longer.

11 years agoUse symbol versioning for our new member of struct pci_dev
Martin Mares [Mon, 1 Apr 2013 13:42:50 +0000 (15:42 +0200)]
Use symbol versioning for our new member of struct pci_dev

11 years agoDetection of module aliases (Linux sysfs only)
Martin Mares [Mon, 1 Apr 2013 13:40:10 +0000 (15:40 +0200)]
Detection of module aliases (Linux sysfs only)

This will be needed to support libkmod.

11 years agoFixed logic error in pci_open()
Martin Mares [Thu, 6 Dec 2012 16:05:15 +0000 (17:05 +0100)]
Fixed logic error in pci_open()

Previously, pci_open() sometimes crashed with ID list names
shorter than 3 characters.