From: Hans de Goede <hdegoede@redhat.com>
To: Shin'ichiro Kawasaki <shinichiro.kawasaki@wdc.com>,
Andy Shevchenko <andy@kernel.org>
Cc: "Ilpo Järvinen" <ilpo.jarvinen@linux.intel.com>,
"Andy Shevchenko" <andriy.shevchenko@linux.intel.com>,
"Lukas Wunner" <lukas@wunner.de>,
linux-pci@vger.kernel.org, linux-i2c@vger.kernel.org,
platform-driver-x86@vger.kernel.org
Subject: Re: [PATCH RFC] platform/x86: p2sb: Allow p2sb_bar() calls during PCI device probe
Date: Mon, 4 Dec 2023 15:10:09 +0100 [thread overview]
Message-ID: <16b251b9-2dbb-419d-bde6-7d5972120587@redhat.com> (raw)
In-Reply-To: <20231201104759.949340-1-shinichiro.kawasaki@wdc.com>
Hi,
On 12/1/23 11:47, Shin'ichiro Kawasaki wrote:
> p2sb_bar() unhides P2SB device to get resources from the device. It
> guards the operation by locking pci_rescan_remove_lock so that parallel
> rescans do not find the P2SB device. However, this lock causes deadlock
> when PCI bus rescan is triggered by /sys/bus/pci/rescan. The rescan
> locks pci_rescan_remove_lock and probes PCI devices. When PCI devices
> call p2sb_bar() during probe, it locks pci_rescan_remove_lock again.
> Hence the deadlock.
>
> To avoid the deadlock, do not lock pci_rescan_remove_lock in p2sb_bar().
> Instead, do the lock at fs_initcall. Introduce p2sb_cache_resources()
> for fs_initcall which gets and caches the P2SB resources. At p2sb_bar(),
> refer the cache and return to the caller.
>
> Link: https://lore.kernel.org/linux-pci/6xb24fjmptxxn5js2fjrrddjae6twex5bjaftwqsuawuqqqydx@7cl3uik5ef6j/
> Suggested-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
> Signed-off-by: Shin'ichiro Kawasaki <shinichiro.kawasaki@wdc.com>
> ---
> This patch reflects discussions held at the Link tag. I confirmed this patch
> fixes the problem using a system with i2c_i801 device, building i2c_i801
> module as both built-in and loadable. Reviews will be appreicated.
Andy, you know this code best, can you review this patch please?
Regards,
Hans
>
> drivers/platform/x86/p2sb.c | 125 +++++++++++++++++++++++++-----------
> 1 file changed, 87 insertions(+), 38 deletions(-)
>
> diff --git a/drivers/platform/x86/p2sb.c b/drivers/platform/x86/p2sb.c
> index 1cf2471d54dd..97e9b44f5f1a 100644
> --- a/drivers/platform/x86/p2sb.c
> +++ b/drivers/platform/x86/p2sb.c
> @@ -26,6 +26,16 @@ static const struct x86_cpu_id p2sb_cpu_ids[] = {
> {}
> };
>
> +/* Cache BAR0 of P2SB device from function 0 ot 7 */
> +#define NR_P2SB_RES_CACHE 8
> +
> +struct p2sb_res_cache {
> + u32 bus_dev_id;
> + struct resource res;
> +};
> +
> +static struct p2sb_res_cache p2sb_resources[NR_P2SB_RES_CACHE];
> +
> static int p2sb_get_devfn(unsigned int *devfn)
> {
> unsigned int fn = P2SB_DEVFN_DEFAULT;
> @@ -39,8 +49,13 @@ static int p2sb_get_devfn(unsigned int *devfn)
> return 0;
> }
>
> +static bool p2sb_invalid_resource(struct resource *res)
> +{
> + return res->flags == 0;
> +}
> +
> /* Copy resource from the first BAR of the device in question */
> -static int p2sb_read_bar0(struct pci_dev *pdev, struct resource *mem)
> +static void p2sb_read_bar0(struct pci_dev *pdev, struct resource *mem)
> {
> struct resource *bar0 = &pdev->resource[0];
>
> @@ -56,48 +71,29 @@ static int p2sb_read_bar0(struct pci_dev *pdev, struct resource *mem)
> mem->end = bar0->end;
> mem->flags = bar0->flags;
> mem->desc = bar0->desc;
> -
> - return 0;
> }
>
> -static int p2sb_scan_and_read(struct pci_bus *bus, unsigned int devfn, struct resource *mem)
> +static int p2sb_scan_and_cache(struct pci_bus *bus, unsigned int devfn)
> {
> struct pci_dev *pdev;
> int ret;
> + struct p2sb_res_cache *cache = &p2sb_resources[PCI_FUNC(devfn)];
>
> pdev = pci_scan_single_device(bus, devfn);
> - if (!pdev)
> + if (!pdev || p2sb_invalid_resource(&pdev->resource[0]))
> return -ENODEV;
>
> - ret = p2sb_read_bar0(pdev, mem);
> + p2sb_read_bar0(pdev, &cache->res);
> + cache->bus_dev_id = bus->dev.id;
>
> pci_stop_and_remove_bus_device(pdev);
> return ret;
> }
>
> -/**
> - * p2sb_bar - Get Primary to Sideband (P2SB) bridge device BAR
> - * @bus: PCI bus to communicate with
> - * @devfn: PCI slot and function to communicate with
> - * @mem: memory resource to be filled in
> - *
> - * The BIOS prevents the P2SB device from being enumerated by the PCI
> - * subsystem, so we need to unhide and hide it back to lookup the BAR.
> - *
> - * if @bus is NULL, the bus 0 in domain 0 will be used.
> - * If @devfn is 0, it will be replaced by devfn of the P2SB device.
> - *
> - * Caller must provide a valid pointer to @mem.
> - *
> - * Locking is handled by pci_rescan_remove_lock mutex.
> - *
> - * Return:
> - * 0 on success or appropriate errno value on error.
> - */
> -int p2sb_bar(struct pci_bus *bus, unsigned int devfn, struct resource *mem)
> +static int p2sb_cache_resources(void)
> {
> - struct pci_dev *pdev_p2sb;
> - unsigned int devfn_p2sb;
> + struct pci_bus *bus;
> + unsigned int devfn_p2sb, slot_p2sb, fn;
> u32 value = P2SBC_HIDE;
> int ret;
>
> @@ -106,8 +102,8 @@ int p2sb_bar(struct pci_bus *bus, unsigned int devfn, struct resource *mem)
> if (ret)
> return ret;
>
> - /* if @bus is NULL, use bus 0 in domain 0 */
> - bus = bus ?: pci_find_bus(0, 0);
> + /* Assume P2SB is on the bus 0 in domain 0 */
> + bus = pci_find_bus(0, 0);
>
> /*
> * Prevent concurrent PCI bus scan from seeing the P2SB device and
> @@ -115,18 +111,31 @@ int p2sb_bar(struct pci_bus *bus, unsigned int devfn, struct resource *mem)
> */
> pci_lock_rescan_remove();
>
> - /* Unhide the P2SB device, if needed */
> + /*
> + * The BIOS prevents the P2SB device from being enumerated by the PCI
> + * subsystem, so we need to unhide and hide it back to lookup the BAR.
> + * Unhide the P2SB device here, if needed.
> + */
> pci_bus_read_config_dword(bus, devfn_p2sb, P2SBC, &value);
> if (value & P2SBC_HIDE)
> pci_bus_write_config_dword(bus, devfn_p2sb, P2SBC, 0);
>
> - pdev_p2sb = pci_scan_single_device(bus, devfn_p2sb);
> - if (devfn)
> - ret = p2sb_scan_and_read(bus, devfn, mem);
> - else
> - ret = p2sb_read_bar0(pdev_p2sb, mem);
> - pci_stop_and_remove_bus_device(pdev_p2sb);
> + /* Scan the P2SB device and cache its BAR0 */
> + ret = p2sb_scan_and_cache(bus, devfn_p2sb);
> + if (ret)
> + goto out;
>
> + /*
> + * When function number of the P2SB device is zero, scan other function
> + * numbers. If devices are available, cache their BAR0.
> + */
> + if (!PCI_FUNC(devfn_p2sb)) {
> + slot_p2sb = PCI_SLOT(devfn_p2sb);
> + for (fn = 1; fn < 8; fn++)
> + p2sb_scan_and_cache(bus, PCI_DEVFN(slot_p2sb, fn));
> + }
> +
> +out:
> /* Hide the P2SB device, if it was hidden */
> if (value & P2SBC_HIDE)
> pci_bus_write_config_dword(bus, devfn_p2sb, P2SBC, P2SBC_HIDE);
> @@ -136,9 +145,49 @@ int p2sb_bar(struct pci_bus *bus, unsigned int devfn, struct resource *mem)
> if (ret)
> return ret;
>
> - if (mem->flags == 0)
> + return 0;
> +}
> +
> +/**
> + * p2sb_bar - Get Primary to Sideband (P2SB) bridge device BAR
> + * @bus: PCI bus to communicate with
> + * @devfn: PCI slot and function to communicate with
> + * @mem: memory resource to be filled in
> + *
> + * if @bus is NULL, the bus 0 in domain 0 will be used.
> + * If @devfn is 0, it will be replaced by devfn of the P2SB device.
> + *
> + * Caller must provide a valid pointer to @mem.
> + *
> + * Return:
> + * 0 on success or appropriate errno value on error.
> + */
> +int p2sb_bar(struct pci_bus *bus, unsigned int devfn, struct resource *mem)
> +{
> + int ret;
> + struct p2sb_res_cache *cache;
> +
> + bus = bus ? bus : pci_find_bus(0, 0);
> +
> + if (!devfn) {
> + ret = p2sb_get_devfn(&devfn);
> + if (ret)
> + return ret;
> + }
> +
> + cache = &p2sb_resources[PCI_FUNC(devfn)];
> + if (p2sb_invalid_resource(&cache->res) ||
> + cache->bus_dev_id != bus->dev.id)
> return -ENODEV;
>
> + memcpy(mem, &cache->res, sizeof(*mem));
> return 0;
> }
> EXPORT_SYMBOL_GPL(p2sb_bar);
> +
> +static int __init p2sb_fs_init(void)
> +{
> + p2sb_cache_resources();
> + return 0;
> +}
> +fs_initcall(p2sb_fs_init);
prev parent reply other threads:[~2023-12-04 14:10 UTC|newest]
Thread overview: 2+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-12-01 10:47 [PATCH RFC] platform/x86: p2sb: Allow p2sb_bar() calls during PCI device probe Shin'ichiro Kawasaki
2023-12-04 14:10 ` Hans de Goede [this message]
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=16b251b9-2dbb-419d-bde6-7d5972120587@redhat.com \
--to=hdegoede@redhat.com \
--cc=andriy.shevchenko@linux.intel.com \
--cc=andy@kernel.org \
--cc=ilpo.jarvinen@linux.intel.com \
--cc=linux-i2c@vger.kernel.org \
--cc=linux-pci@vger.kernel.org \
--cc=lukas@wunner.de \
--cc=platform-driver-x86@vger.kernel.org \
--cc=shinichiro.kawasaki@wdc.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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox