From: Bjorn Helgaas <bhelgaas@google.com>
To: Yinghai Lu <yinghai@kernel.org>
Cc: "Marek Kordík" <kordikmarek@gmail.com>,
"Gavin Shan" <gwshan@linux.vnet.ibm.com>,
"Benjamin Herrenschmidt" <benh@kernel.crashing.org>,
"linux-pci@vger.kernel.org" <linux-pci@vger.kernel.org>,
"Alexey Voronkov" <zermond@gmail.com>,
"David Airlie" <airlied@linux.ie>,
"Alex Deucher" <alexdeucher@gmail.com>,
"Richard Yang" <weiyang@linux.vnet.ibm.com>
Subject: Re: Regression: bug 85491: radeon 0000:01:00.0: Fatal error during GPU init
Date: Fri, 9 Jan 2015 11:45:49 -0700 [thread overview]
Message-ID: <20150109184549.GI6575@google.com> (raw)
In-Reply-To: <CAE9FiQUihEJeMT98pF-eQTEY9NcvvBCUgWQKXVXwmCZ8=Jy=8Q@mail.gmail.com>
On Thu, Dec 25, 2014 at 01:12:47PM -0800, Yinghai Lu wrote:
> On Thu, Dec 25, 2014 at 7:46 AM, Marek Kordík <kordikmarek@gmail.com> wrote:
> > On 12/24/2014 03:29 AM, Yinghai Lu wrote:
>
> > I checked (only) v2 of your patch on versions 3.18.0 and 3.18.1. The result
> > of both versions is that booting stops with kernel panic. I don't how to get
> > the full log (there is nothing in journalctl and dmesg after booting some
> > working kernel - maybe if you can give me some hint what to google) so I
> > will attach photo of screen to the bugzilla issue.
>
> Please check update one.
>
> but i still can not understand why do we need
>
> if (!r)
> continue;
>
> checking...
> Subject: [PATCH v3] PCI, x86: clip firmware assigned pci bridges under hostbridge
>
> Some bios put range that is not fully coverred by root bus resources.
> Try to clip them and update them in pci bridge bars.
>
> -v2: check with upstream bridge instead of host bridge only.
> also add debug print out.
> -v3: fix checking under pci_bus_for_each...
>
> Link: https://bugzilla.kernel.org/show_bug.cgi?id=85491
> Reported-by: Marek Kordik <kordikmarek@gmail.com>
> Fixes: 5b28541552ef ("PCI: Restrict 64-bit prefetchable bridge windows to 64-bit resources")
> Signed-off-by: Yinghai Lu <yinghai@kernel.org>
This is becoming urgent.
Do we have a patch that fixes the problem on x86? I see that v2 failed for
Marek, and I don't see any test results for v3.
Also, my intent is for Marek or others to test this and verify that it
works on x86, and then you extend the patch to cover all architectures, and
then I'll merge them all. I do not intend to merge the x86 change by
itself before the changes to other arches, because if I do that, the other
arches will probably never get fixed.
My only other option is to revert 5b28541552ef. That would break stuff on
powerpc, so I don't want to do it, but we need to resolve this soon.
Bjorn
> ---
> arch/x86/pci/i386.c | 81 +++++++++++++++++++++++++++++++++++++---------------
> drivers/pci/bus.c | 37 +++++++++++++++++++++++
> include/linux/pci.h | 1
> 3 files changed, 97 insertions(+), 22 deletions(-)
>
> Index: linux-2.6/arch/x86/pci/i386.c
> ===================================================================
> --- linux-2.6.orig/arch/x86/pci/i386.c
> +++ linux-2.6/arch/x86/pci/i386.c
> @@ -205,28 +205,48 @@ EXPORT_SYMBOL(pcibios_align_resource);
> * as well.
> */
>
> -static void pcibios_allocate_bridge_resources(struct pci_dev *dev)
> +static bool pcibios_allocate_bridge_resources(struct pci_dev *dev)
> {
> int idx;
> struct resource *r;
> + bool changed = false;
>
> for (idx = PCI_BRIDGE_RESOURCES; idx < PCI_NUM_RESOURCES; idx++) {
> + int ret;
> +
> r = &dev->resource[idx];
> if (!r->flags)
> continue;
> if (r->parent) /* Already allocated */
> continue;
> - if (!r->start || pci_claim_resource(dev, idx) < 0) {
> - /*
> - * Something is wrong with the region.
> - * Invalidate the resource to prevent
> - * child resource allocations in this
> - * range.
> - */
> - r->start = r->end = 0;
> - r->flags = 0;
> +
> + if (!r->start)
> + goto clear;
> +
> + ret = pci_claim_resource(dev, idx);
> + if (ret >= 0)
> + continue;
> +
> + /* try again after clip for pci bridge*/
> + if ((dev->class >> 8) == PCI_CLASS_BRIDGE_PCI &&
> + pci_bus_clip_resource(dev, r)) {
> + changed = true;
> + if (pci_claim_resource(dev, idx) >= 0)
> + continue;
> }
> +
> +clear:
> + /*
> + * Something is wrong with the region.
> + * Invalidate the resource to prevent
> + * child resource allocations in this
> + * range.
> + */
> + r->start = r->end = 0;
> + r->flags = 0;
> }
> +
> + return changed;
> }
>
> static void pcibios_allocate_bus_resources(struct pci_bus *bus)
> @@ -234,8 +254,12 @@ static void pcibios_allocate_bus_resourc
> struct pci_bus *child;
>
> /* Depth-First Search on bus tree */
> - if (bus->self)
> - pcibios_allocate_bridge_resources(bus->self);
> + if (bus->self) {
> + bool changed = pcibios_allocate_bridge_resources(bus->self);
> +
> + if (changed)
> + pci_setup_bridge(bus);
> + }
> list_for_each_entry(child, &bus->children, node)
> pcibios_allocate_bus_resources(child);
> }
> @@ -271,21 +295,34 @@ static void pcibios_allocate_dev_resourc
> else
> disabled = !(command & PCI_COMMAND_MEMORY);
> if (pass == disabled) {
> + int ret;
> +
> dev_dbg(&dev->dev,
> "BAR %d: reserving %pr (d=%d, p=%d)\n",
> idx, r, disabled, pass);
> - if (pci_claim_resource(dev, idx) < 0) {
> - if (r->flags & IORESOURCE_PCI_FIXED) {
> - dev_info(&dev->dev, "BAR %d %pR is immovable\n",
> +
> + ret = pci_claim_resource(dev, idx);
> +
> + if (ret >= 0)
> + continue;
> +
> + if (r->flags & IORESOURCE_PCI_FIXED) {
> + dev_info(&dev->dev, "BAR %d %pR is immovable\n",
> idx, r);
> - } else {
> - /* We'll assign a new address later */
> - pcibios_save_fw_addr(dev,
> - idx, r->start);
> - r->end -= r->start;
> - r->start = 0;
> - }
> + continue;
> + }
> +
> + /* try again with clip */
> + if (pci_bus_clip_resource(dev, r)) {
> + pci_update_resource(dev, idx);
> + if (pci_claim_resource(dev, idx) >= 0)
> + continue;
> }
> +
> + /* We'll assign a new address later */
> + pcibios_save_fw_addr(dev, idx, r->start);
> + r->end -= r->start;
> + r->start = 0;
> }
> }
> if (!pass) {
> Index: linux-2.6/include/linux/pci.h
> ===================================================================
> --- linux-2.6.orig/include/linux/pci.h
> +++ linux-2.6/include/linux/pci.h
> @@ -1094,6 +1094,7 @@ void pci_free_resource_list(struct list_
> void pci_bus_add_resource(struct pci_bus *bus, struct resource *res, unsigned int flags);
> struct resource *pci_bus_resource_n(const struct pci_bus *bus, int n);
> void pci_bus_remove_resources(struct pci_bus *bus);
> +bool pci_bus_clip_resource(struct pci_dev *dev, struct resource *res);
>
> #define pci_bus_for_each_resource(bus, res, i) \
> for (i = 0; \
> Index: linux-2.6/drivers/pci/bus.c
> ===================================================================
> --- linux-2.6.orig/drivers/pci/bus.c
> +++ linux-2.6/drivers/pci/bus.c
> @@ -228,6 +228,43 @@ int pci_bus_alloc_resource(struct pci_bu
> }
> EXPORT_SYMBOL(pci_bus_alloc_resource);
>
> +bool pci_bus_clip_resource(struct pci_dev *dev, struct resource *res)
> +{
> + struct pci_bus *bus = dev->bus;
> + resource_size_t start, end;
> + struct resource orig_res = *res;
> + struct resource *r;
> + int i;
> +
> + pci_bus_for_each_resource(bus, r, i) {
> + if (!r)
> + continue;
> +
> + if (resource_type(res) != resource_type(r))
> + continue;
> +
> + start = max(r->start, res->start);
> + end = min(r->end, res->end);
> +
> + /* no overlap ? */
> + if (start > end)
> + continue;
> +
> + if (res->start == start && res->end == end)
> + return false;
> +
> + /* changed */
> + res->start = start;
> + res->end = end;
> + dev_printk(KERN_DEBUG, &dev->dev, "%pR ==> %pR\n",
> + &orig_res, res);
> +
> + return true;
> + }
> +
> + return false;
> +}
> +
> void __weak pcibios_resource_survey_bus(struct pci_bus *bus) { }
>
> /**
next prev parent reply other threads:[~2015-01-09 18:45 UTC|newest]
Thread overview: 37+ messages / expand[flat|nested] mbox.gz Atom feed top
2014-12-03 22:15 Regression: bug 85491: radeon 0000:01:00.0: Fatal error during GPU init Bjorn Helgaas
2014-12-04 0:51 ` Bjorn Helgaas
2014-12-04 1:44 ` Yinghai Lu
2014-12-04 17:34 ` Bjorn Helgaas
2014-12-04 22:06 ` Yinghai Lu
2014-12-05 10:59 ` Benjamin Herrenschmidt
2014-12-08 0:47 ` Gavin Shan
2014-12-08 21:04 ` Bjorn Helgaas
2014-12-08 21:38 ` Benjamin Herrenschmidt
2014-12-08 21:46 ` Bjorn Helgaas
2014-12-08 23:38 ` Gavin Shan
2014-12-09 0:21 ` Yinghai Lu
2014-12-19 22:17 ` Bjorn Helgaas
2014-12-20 0:23 ` Bjorn Helgaas
2014-12-20 0:34 ` Yinghai Lu
2014-12-20 0:56 ` Bjorn Helgaas
2014-12-20 1:33 ` Yinghai Lu
2014-12-23 1:53 ` Yinghai Lu
2014-12-23 21:41 ` Bjorn Helgaas
2014-12-24 2:29 ` Yinghai Lu
2014-12-24 17:26 ` Bjorn Helgaas
2014-12-24 20:29 ` Yinghai Lu
2014-12-25 15:46 ` Marek Kordík
2014-12-25 21:12 ` Yinghai Lu
2015-01-09 18:45 ` Bjorn Helgaas [this message]
2015-01-09 20:38 ` Yinghai Lu
2014-12-22 8:17 ` Wei Yang
2014-12-22 19:50 ` Bjorn Helgaas
2014-12-04 1:38 ` Yinghai Lu
2014-12-04 1:41 ` Yinghai Lu
2014-12-04 16:15 ` Bjorn Helgaas
2014-12-04 22:15 ` Yinghai Lu
2014-12-04 22:25 ` Bjorn Helgaas
2014-12-04 2:01 ` Yinghai Lu
2014-12-04 16:37 ` Bjorn Helgaas
2014-12-04 22:24 ` Yinghai Lu
2014-12-05 0:24 ` Yinghai Lu
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20150109184549.GI6575@google.com \
--to=bhelgaas@google.com \
--cc=airlied@linux.ie \
--cc=alexdeucher@gmail.com \
--cc=benh@kernel.crashing.org \
--cc=gwshan@linux.vnet.ibm.com \
--cc=kordikmarek@gmail.com \
--cc=linux-pci@vger.kernel.org \
--cc=weiyang@linux.vnet.ibm.com \
--cc=yinghai@kernel.org \
--cc=zermond@gmail.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.