Can we get 1f311c94aabdb419c28e3147bcc8ab89269f1a7e merged into the stable tree?
I have some compatibility issues on Realtek chips because of the missing initialization clocks.
Thanks!
Regards,
Christian
From: Ulf Hansson <ulf.hansson(a)linaro.org>
Sent: Monday, February 28, 2022 5:12 PM
To: Ricky WU
Cc: gregkh(a)linuxfoundation.org; kai.heng.feng(a)canonical.com; tommyhebb(a)gmail.com; linux-mmc(a)vger.kernel.org; linux-kernel(a)vger.kernel.org
Subject: Re: [PATCH] mmc: rtsx: add 74 Clocks in power on flow
On Tue, 22 Feb 2022 at 08:28, Ricky WU <ricky_wu(a)realtek.com> wrote:
>
> After 1ms stabilizing the voltage time
> add "Host provides at least 74 Clocks
> before issuing first command" that is
> spec definition
>
> Signed-off-by: Ricky Wu <ricky_wu(a)realtek.com>
> ---
> drivers/mmc/host/rtsx_pci_sdmmc.c | 7 +++++++
> 1 file changed, 7 insertions(+)
>
> diff --git a/drivers/mmc/host/rtsx_pci_sdmmc.c b/drivers/mmc/host/rtsx_pci_sdmmc.c
> index 2a3f14afe9f8..e016d720e453 100644
> --- a/drivers/mmc/host/rtsx_pci_sdmmc.c
> +++ b/drivers/mmc/host/rtsx_pci_sdmmc.c
> @@ -940,10 +940,17 @@ static int sd_power_on(struct realtek_pci_sdmmc *host)
> if (err < 0)
> return err;
>
> + mdelay(1);
> +
> err = rtsx_pci_write_register(pcr, CARD_OE, SD_OUTPUT_EN, SD_OUTPUT_EN);
> if (err < 0)
> return err;
>
> + /* send init 74 clocks */
> + rtsx_pci_write_register(pcr, SD_BUS_STAT, SD_CLK_TOGGLE_EN, SD_CLK_TOGGLE_EN);
> + mdelay(5);
> + rtsx_pci_write_register(pcr, SD_BUS_STAT, SD_CLK_TOGGLE_EN, 0);
> +
> if (PCI_PID(pcr) == PID_5261) {
> /*
> * If test mode is set switch to SD Express mandatorily,
As you probably are aware of, the mmc core uses three power states
(MMC_POWER_ON, MMC_POWER_UP and MMC_POWER_OFF) to manage the
initialization, while it invokes the ->set_ios() callback for the mmc
host driver. During these steps the core also tries to manage the
different delays that are needed according to the eMMC/SD specs. You
may have a look at mmc_power_up() in drivers/mmc/core/core.c. In the
rtsx case, MMC_POWER_ON and MMC_POWER_UP are treated as one single
step.
Moreover, it has turned out that some mmc HWs are actually controlling
these delays during the initialization themselves, which makes the
delays in the core superfluous. Therefore we have made the delays
configurable for host drivers. For DT based platforms, we have the DT
property "post-power-on-delay-ms" and for others, it's perfectly fine
to update host->power_delay_ms before calling mmc_add_host().
Would it be possible to take advantage of the above "features" from
the core, to avoid hard coded and superfluous delays?
Kind regards
Uffe
=
Hyperstone GmbH | Reichenaustr. 39a | 78467 Konstanz
Managing Director: Dr. Jan Peter Berns.
Commercial register of local courts: Freiburg HRB381782
We have run into an issue that a task gets stuck in
balance_dirty_pages_ratelimited() when perform I/O stress testing.
The reason we observed is that an I_DIRTY_PAGES inode with lots
of dirty pages is in b_dirty_time list and standard background
writeback cannot writeback the inode.
After studing the relevant code, the following scenario may lead
to the issue:
task1 task2
----- -----
fuse_flush
write_inode_now //in b_dirty_time
writeback_single_inode
__writeback_single_inode
fuse_write_end
filemap_dirty_folio
__xa_set_mark:PAGECACHE_TAG_DIRTY
lock inode->i_lock
if mapping tagged PAGECACHE_TAG_DIRTY
inode->i_state |= I_DIRTY_PAGES
unlock inode->i_lock
__mark_inode_dirty:I_DIRTY_PAGES
lock inode->i_lock
-was dirty,inode stays in
-b_dirty_time
unlock inode->i_lock
if(!(inode->i_state & I_DIRTY_All))
-not true,so nothing done
This patch moves the dirty inode to b_dirty list when the inode
currently is not queued in b_io or b_more_io list at the end of
writeback_single_inode.
Reviewed-by: Jan Kara <jack(a)suse.cz>
Reviewed-by: Christoph Hellwig <hch(a)lst.de>
CC: stable(a)vger.kernel.org
Fixes: 0ae45f63d4ef ("vfs: add support for a lazytime mount option")
Signed-off-by: Jing Xia <jing.xia(a)unisoc.com>
---
fs/fs-writeback.c | 4 ++++
1 file changed, 4 insertions(+)
diff --git a/fs/fs-writeback.c b/fs/fs-writeback.c
index 591fe9cf1659..1fae0196292a 100644
--- a/fs/fs-writeback.c
+++ b/fs/fs-writeback.c
@@ -1712,6 +1712,10 @@ static int writeback_single_inode(struct inode *inode,
*/
if (!(inode->i_state & I_DIRTY_ALL))
inode_cgwb_move_to_attached(inode, wb);
+ else if (!(inode->i_state & I_SYNC_QUEUED) &&
+ (inode->i_state & I_DIRTY))
+ redirty_tail_locked(inode, wb);
+
spin_unlock(&wb->list_lock);
inode_sync_complete(inode);
out:
--
2.17.1
Set "HCD_FLAG_DEFER_RH_REGISTER" to hcd->flags in xhci_run() to defer
registering primary roothub in usb_add_hcd() if xhci has two roothubs.
This will make sure both primary roothub and secondary roothub will be
registered along with the second HCD.
This is required for cold plugged USB devices to be detected in certain
PCIe USB cards (like Inateck USB card connected to AM64 EVM or J7200 EVM).
This patch has been added and reverted earier as it triggered a race
in usb device enumeration.
That race is now fixed in 5.16-rc3, and in stable back to 5.4
commit 6cca13de26ee ("usb: hub: Fix locking issues with address0_mutex")
commit 6ae6dc22d2d1 ("usb: hub: Fix usb enumeration issue due to address0
race")
[minor rebase change, and commit message update -Mathias]
CC: stable(a)vger.kernel.org # 5.4+
Suggested-by: Mathias Nyman <mathias.nyman(a)linux.intel.com>
Tested-by: Chris Chiu <chris.chiu(a)canonical.com>
Signed-off-by: Kishon Vijay Abraham I <kishon(a)ti.com>
---
drivers/usb/host/xhci.c | 2 ++
1 file changed, 2 insertions(+)
diff --git a/drivers/usb/host/xhci.c b/drivers/usb/host/xhci.c
index 25b87e99b4dd..2be38d9de8df 100644
--- a/drivers/usb/host/xhci.c
+++ b/drivers/usb/host/xhci.c
@@ -696,6 +696,8 @@ int xhci_run(struct usb_hcd *hcd)
xhci_dbg_trace(xhci, trace_xhci_dbg_init,
"Finished xhci_run for USB2 roothub");
+ set_bit(HCD_FLAG_DEFER_RH_REGISTER, &hcd->flags);
+
xhci_create_dbc_dev(xhci);
xhci_debugfs_init(xhci);
--
2.17.1
It has been observed with certain PCIe USB cards (like Inateck connected
to AM64 EVM or J7200 EVM) that as soon as the primary roothub is
registered, port status change is handled even before xHC is running
leading to cold plug USB devices not detected. For such cases, registering
both the root hubs along with the second HCD is required. Add support for
deferring roothub registration in usb_add_hcd(), so that both primary and
secondary roothubs are registered along with the second HCD.
This patch has been added and reverted earier as it triggered a race
in usb device enumeration.
That race is now fixed in 5.16-rc3, and in stable back to 5.4
commit 6cca13de26ee ("usb: hub: Fix locking issues with address0_mutex")
commit 6ae6dc22d2d1 ("usb: hub: Fix usb enumeration issue due to address0
race")
CC: stable(a)vger.kernel.org # 5.4+
Suggested-by: Mathias Nyman <mathias.nyman(a)linux.intel.com>
Tested-by: Chris Chiu <chris.chiu(a)canonical.com>
Acked-by: Alan Stern <stern(a)rowland.harvard.edu>
Signed-off-by: Kishon Vijay Abraham I <kishon(a)ti.com>
---
drivers/usb/core/hcd.c | 29 +++++++++++++++++++++++------
include/linux/usb/hcd.h | 2 ++
2 files changed, 25 insertions(+), 6 deletions(-)
diff --git a/drivers/usb/core/hcd.c b/drivers/usb/core/hcd.c
index d9712c2602af..06eea8848ccc 100644
--- a/drivers/usb/core/hcd.c
+++ b/drivers/usb/core/hcd.c
@@ -2816,6 +2816,7 @@ int usb_add_hcd(struct usb_hcd *hcd,
{
int retval;
struct usb_device *rhdev;
+ struct usb_hcd *shared_hcd;
if (!hcd->skip_phy_initialization && usb_hcd_is_primary_hcd(hcd)) {
hcd->phy_roothub = usb_phy_roothub_alloc(hcd->self.sysdev);
@@ -2976,13 +2977,26 @@ int usb_add_hcd(struct usb_hcd *hcd,
goto err_hcd_driver_start;
}
+ /* starting here, usbcore will pay attention to the shared HCD roothub */
+ shared_hcd = hcd->shared_hcd;
+ if (!usb_hcd_is_primary_hcd(hcd) && shared_hcd && HCD_DEFER_RH_REGISTER(shared_hcd)) {
+ retval = register_root_hub(shared_hcd);
+ if (retval != 0)
+ goto err_register_root_hub;
+
+ if (shared_hcd->uses_new_polling && HCD_POLL_RH(shared_hcd))
+ usb_hcd_poll_rh_status(shared_hcd);
+ }
+
/* starting here, usbcore will pay attention to this root hub */
- retval = register_root_hub(hcd);
- if (retval != 0)
- goto err_register_root_hub;
+ if (!HCD_DEFER_RH_REGISTER(hcd)) {
+ retval = register_root_hub(hcd);
+ if (retval != 0)
+ goto err_register_root_hub;
- if (hcd->uses_new_polling && HCD_POLL_RH(hcd))
- usb_hcd_poll_rh_status(hcd);
+ if (hcd->uses_new_polling && HCD_POLL_RH(hcd))
+ usb_hcd_poll_rh_status(hcd);
+ }
return retval;
@@ -3020,6 +3034,7 @@ EXPORT_SYMBOL_GPL(usb_add_hcd);
void usb_remove_hcd(struct usb_hcd *hcd)
{
struct usb_device *rhdev = hcd->self.root_hub;
+ bool rh_registered;
dev_info(hcd->self.controller, "remove, state %x\n", hcd->state);
@@ -3030,6 +3045,7 @@ void usb_remove_hcd(struct usb_hcd *hcd)
dev_dbg(hcd->self.controller, "roothub graceful disconnect\n");
spin_lock_irq (&hcd_root_hub_lock);
+ rh_registered = hcd->rh_registered;
hcd->rh_registered = 0;
spin_unlock_irq (&hcd_root_hub_lock);
@@ -3039,7 +3055,8 @@ void usb_remove_hcd(struct usb_hcd *hcd)
cancel_work_sync(&hcd->died_work);
mutex_lock(&usb_bus_idr_lock);
- usb_disconnect(&rhdev); /* Sets rhdev to NULL */
+ if (rh_registered)
+ usb_disconnect(&rhdev); /* Sets rhdev to NULL */
mutex_unlock(&usb_bus_idr_lock);
/*
diff --git a/include/linux/usb/hcd.h b/include/linux/usb/hcd.h
index 548a028f2dab..2c1fc9212cf2 100644
--- a/include/linux/usb/hcd.h
+++ b/include/linux/usb/hcd.h
@@ -124,6 +124,7 @@ struct usb_hcd {
#define HCD_FLAG_RH_RUNNING 5 /* root hub is running? */
#define HCD_FLAG_DEAD 6 /* controller has died? */
#define HCD_FLAG_INTF_AUTHORIZED 7 /* authorize interfaces? */
+#define HCD_FLAG_DEFER_RH_REGISTER 8 /* Defer roothub registration */
/* The flags can be tested using these macros; they are likely to
* be slightly faster than test_bit().
@@ -134,6 +135,7 @@ struct usb_hcd {
#define HCD_WAKEUP_PENDING(hcd) ((hcd)->flags & (1U << HCD_FLAG_WAKEUP_PENDING))
#define HCD_RH_RUNNING(hcd) ((hcd)->flags & (1U << HCD_FLAG_RH_RUNNING))
#define HCD_DEAD(hcd) ((hcd)->flags & (1U << HCD_FLAG_DEAD))
+#define HCD_DEFER_RH_REGISTER(hcd) ((hcd)->flags & (1U << HCD_FLAG_DEFER_RH_REGISTER))
/*
* Specifies if interfaces are authorized by default
--
2.17.1