| Message ID | 1307367739-20110-1-git-send-email-keshava_mgowda@ti.com (mailing list archive) |
|---|---|
| State | New, archived |
| Delegated to: | Felipe Balbi |
| Headers | show |
On Mon, Jun 06, 2011 at 07:12:19PM +0530, Keshava Munegowda wrote: > From: Keshava Munegowda <Keshava_mgowda@ti.com> > > Oops are produced during initialization of ehci and ohci > drivers. This is because the run time pm apis are used by > the driver but the corresponding hwmod structures and > initialization is not merged. hence revering back the > commit id 7e6502d577106fb5b202bbaac64c5f1b065e6daa > > Signed-off-by: Keshava Munegowda <keshava_mgowda@ti.com> Reported-by: Luciano Coelho <coelho@ti.com> Acked-by: Felipe Balbi <balbi@ti.com>
Why is this not just a straight revert of 7e6502d577106fb5b202bbaac64c5f1b065e6daa? On Mon, Jun 6, 2011 at 6:52 AM, Felipe Balbi <balbi@ti.com> wrote: > On Mon, Jun 06, 2011 at 07:12:19PM +0530, Keshava Munegowda wrote: >> From: Keshava Munegowda <Keshava_mgowda@ti.com> >> >> Oops are produced during initialization of ehci and ohci >> drivers. This is because the run time pm apis are used by >> the driver but the corresponding hwmod structures and >> initialization is not merged. hence revering back the >> commit id 7e6502d577106fb5b202bbaac64c5f1b065e6daa >> >> Signed-off-by: Keshava Munegowda <keshava_mgowda@ti.com> > > Reported-by: Luciano Coelho <coelho@ti.com> > Acked-by: Felipe Balbi <balbi@ti.com> > > -- > balbi > -- To unsubscribe from this list: send the line "unsubscribe linux-omap" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Hi Keshava, On Mon, Jun 06, 2011 at 07:12:19PM +0530, Keshava Munegowda wrote: > From: Keshava Munegowda <Keshava_mgowda@ti.com> > > Oops are produced during initialization of ehci and ohci > drivers. This is because the run time pm apis are used by > the driver but the corresponding hwmod structures and > initialization is not merged. You mean they're currently checked in a different tree ? Is that a public one? Cheers, Samuel.
Hi, On Mon, Jun 20, 2011 at 03:26:26PM +0200, Samuel Ortiz wrote: > On Mon, Jun 06, 2011 at 07:12:19PM +0530, Keshava Munegowda wrote: > > From: Keshava Munegowda <Keshava_mgowda@ti.com> > > > > Oops are produced during initialization of ehci and ohci > > drivers. This is because the run time pm apis are used by > > the driver but the corresponding hwmod structures and > > initialization is not merged. > You mean they're currently checked in a different tree ? Is that a > public one? it was supposed to go via linux-omap tree but the patches got lost in the limbo :-(
Hi Felipe, On Mon, Jun 20, 2011 at 04:28:52PM +0300, Felipe Balbi wrote: > Hi, > > On Mon, Jun 20, 2011 at 03:26:26PM +0200, Samuel Ortiz wrote: > > On Mon, Jun 06, 2011 at 07:12:19PM +0530, Keshava Munegowda wrote: > > > From: Keshava Munegowda <Keshava_mgowda@ti.com> > > > > > > Oops are produced during initialization of ehci and ohci > > > drivers. This is because the run time pm apis are used by > > > the driver but the corresponding hwmod structures and > > > initialization is not merged. > > You mean they're currently checked in a different tree ? Is that a > > public one? > > it was supposed to go via linux-omap tree but the patches got lost in > the limbo :-( Then shouldn't those patches be the ones to be sent to Linus as a fix for 3.0 ? Cheers, Samuel.
Samuel Ortiz <sameo@linux.intel.com> writes: > Hi Felipe, > > On Mon, Jun 20, 2011 at 04:28:52PM +0300, Felipe Balbi wrote: >> Hi, >> >> On Mon, Jun 20, 2011 at 03:26:26PM +0200, Samuel Ortiz wrote: >> > On Mon, Jun 06, 2011 at 07:12:19PM +0530, Keshava Munegowda wrote: >> > > From: Keshava Munegowda <Keshava_mgowda@ti.com> >> > > >> > > Oops are produced during initialization of ehci and ohci >> > > drivers. This is because the run time pm apis are used by >> > > the driver but the corresponding hwmod structures and >> > > initialization is not merged. >> > You mean they're currently checked in a different tree ? Is that a >> > public one? >> >> it was supposed to go via linux-omap tree but the patches got lost in >> the limbo :-( > Then shouldn't those patches be the ones to be sent to Linus as a fix for 3.0 ? If they were ready, maybe. But those patches still need important work (and review) and are not "fix" material but need to wait until the next merge window. Basically, the original patch should not have been submitted to mainline until the runtime PM support was ready, so the correct short term fix is to simply revert. Also, to echo the question from Dima Zavin: Why isn't this just a simple revert of the original patch? Kevin -- To unsubscribe from this list: send the line "unsubscribe linux-omap" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Hi, On Mon, Jun 20, 2011 at 03:06:01PM -0700, Kevin Hilman wrote: > Samuel Ortiz <sameo@linux.intel.com> writes: > > > Hi Felipe, > > > > On Mon, Jun 20, 2011 at 04:28:52PM +0300, Felipe Balbi wrote: > >> Hi, > >> > >> On Mon, Jun 20, 2011 at 03:26:26PM +0200, Samuel Ortiz wrote: > >> > On Mon, Jun 06, 2011 at 07:12:19PM +0530, Keshava Munegowda wrote: > >> > > From: Keshava Munegowda <Keshava_mgowda@ti.com> > >> > > > >> > > Oops are produced during initialization of ehci and ohci > >> > > drivers. This is because the run time pm apis are used by > >> > > the driver but the corresponding hwmod structures and > >> > > initialization is not merged. > >> > You mean they're currently checked in a different tree ? Is that a > >> > public one? > >> > >> it was supposed to go via linux-omap tree but the patches got lost in > >> the limbo :-( > > Then shouldn't those patches be the ones to be sent to Linus as a fix for 3.0 ? > > If they were ready, maybe. But those patches still need important work > (and review) and are not "fix" material but need to wait until the next > merge window. > > Basically, the original patch should not have been submitted to mainline > until the runtime PM support was ready, so the correct short term fix is > to simply revert. > > Also, to echo the question from Dima Zavin: > > Why isn't this just a simple revert of the original patch? good question. git revert 7e6502d577106fb5b202bbaac64c5f1b065e6daa is much better.
Hi Kevin, On Mon, Jun 20, 2011 at 03:06:01PM -0700, Kevin Hilman wrote: > Samuel Ortiz <sameo@linux.intel.com> writes: > > > Hi Felipe, > > > > On Mon, Jun 20, 2011 at 04:28:52PM +0300, Felipe Balbi wrote: > >> Hi, > >> > >> On Mon, Jun 20, 2011 at 03:26:26PM +0200, Samuel Ortiz wrote: > >> > On Mon, Jun 06, 2011 at 07:12:19PM +0530, Keshava Munegowda wrote: > >> > > From: Keshava Munegowda <Keshava_mgowda@ti.com> > >> > > > >> > > Oops are produced during initialization of ehci and ohci > >> > > drivers. This is because the run time pm apis are used by > >> > > the driver but the corresponding hwmod structures and > >> > > initialization is not merged. > >> > You mean they're currently checked in a different tree ? Is that a > >> > public one? > >> > >> it was supposed to go via linux-omap tree but the patches got lost in > >> the limbo :-( > > Then shouldn't those patches be the ones to be sent to Linus as a fix for 3.0 ? > > If they were ready, maybe. But those patches still need important work > (and review) and are not "fix" material but need to wait until the next > merge window. Fair enough. > Basically, the original patch should not have been submitted to mainline > until the runtime PM support was ready, so the correct short term fix is > to simply revert. > > Also, to echo the question from Dima Zavin: > > Why isn't this just a simple revert of the original patch? I did a revert in my tree. Cheers, Samuel.
Samuel Ortiz <sameo@linux.intel.com> writes: > Hi Kevin, > > On Mon, Jun 20, 2011 at 03:06:01PM -0700, Kevin Hilman wrote: >> Samuel Ortiz <sameo@linux.intel.com> writes: >> >> > Hi Felipe, >> > >> > On Mon, Jun 20, 2011 at 04:28:52PM +0300, Felipe Balbi wrote: >> >> Hi, >> >> >> >> On Mon, Jun 20, 2011 at 03:26:26PM +0200, Samuel Ortiz wrote: >> >> > On Mon, Jun 06, 2011 at 07:12:19PM +0530, Keshava Munegowda wrote: >> >> > > From: Keshava Munegowda <Keshava_mgowda@ti.com> >> >> > > >> >> > > Oops are produced during initialization of ehci and ohci >> >> > > drivers. This is because the run time pm apis are used by >> >> > > the driver but the corresponding hwmod structures and >> >> > > initialization is not merged. >> >> > You mean they're currently checked in a different tree ? Is that a >> >> > public one? >> >> >> >> it was supposed to go via linux-omap tree but the patches got lost in >> >> the limbo :-( >> > Then shouldn't those patches be the ones to be sent to Linus as a fix for 3.0 ? >> >> If they were ready, maybe. But those patches still need important work >> (and review) and are not "fix" material but need to wait until the next >> merge window. > Fair enough. > > >> Basically, the original patch should not have been submitted to mainline >> until the runtime PM support was ready, so the correct short term fix is >> to simply revert. >> >> Also, to echo the question from Dima Zavin: >> >> Why isn't this just a simple revert of the original patch? > I did a revert in my tree. Great, that's the cleanest solution IMO. Thanks! Kevin -- To unsubscribe from this list: send the line "unsubscribe linux-omap" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
diff --git a/drivers/mfd/omap-usb-host.c b/drivers/mfd/omap-usb-host.c index 8552195..ab879e5 100644 --- a/drivers/mfd/omap-usb-host.c +++ b/drivers/mfd/omap-usb-host.c @@ -26,7 +26,6 @@ #include <linux/spinlock.h> #include <linux/gpio.h> #include <plat/usb.h> -#include <linux/pm_runtime.h> #define USBHS_DRIVER_NAME "usbhs-omap" #define OMAP_EHCI_DEVICE "ehci-omap" @@ -147,6 +146,9 @@ struct usbhs_hcd_omap { + struct clk *usbhost_ick; + struct clk *usbhost_hs_fck; + struct clk *usbhost_fs_fck; struct clk *xclk60mhsp1_ck; struct clk *xclk60mhsp2_ck; struct clk *utmi_p1_fck; @@ -156,6 +158,8 @@ struct usbhs_hcd_omap { struct clk *usbhost_p2_fck; struct clk *usbtll_p2_fck; struct clk *init_60m_fclk; + struct clk *usbtll_fck; + struct clk *usbtll_ick; void __iomem *uhh_base; void __iomem *tll_base; @@ -349,13 +353,46 @@ static int __devinit usbhs_omap_probe(struct platform_device *pdev) omap->platdata.ehci_data = pdata->ehci_data; omap->platdata.ohci_data = pdata->ohci_data; - pm_runtime_enable(&pdev->dev); + omap->usbhost_ick = clk_get(dev, "usbhost_ick"); + if (IS_ERR(omap->usbhost_ick)) { + ret = PTR_ERR(omap->usbhost_ick); + dev_err(dev, "usbhost_ick failed error:%d\n", ret); + goto err_end; + } + + omap->usbhost_hs_fck = clk_get(dev, "hs_fck"); + if (IS_ERR(omap->usbhost_hs_fck)) { + ret = PTR_ERR(omap->usbhost_hs_fck); + dev_err(dev, "usbhost_hs_fck failed error:%d\n", ret); + goto err_usbhost_ick; + } + + omap->usbhost_fs_fck = clk_get(dev, "fs_fck"); + if (IS_ERR(omap->usbhost_fs_fck)) { + ret = PTR_ERR(omap->usbhost_fs_fck); + dev_err(dev, "usbhost_fs_fck failed error:%d\n", ret); + goto err_usbhost_hs_fck; + } + + omap->usbtll_fck = clk_get(dev, "usbtll_fck"); + if (IS_ERR(omap->usbtll_fck)) { + ret = PTR_ERR(omap->usbtll_fck); + dev_err(dev, "usbtll_fck failed error:%d\n", ret); + goto err_usbhost_fs_fck; + } + + omap->usbtll_ick = clk_get(dev, "usbtll_ick"); + if (IS_ERR(omap->usbtll_ick)) { + ret = PTR_ERR(omap->usbtll_ick); + dev_err(dev, "usbtll_ick failed error:%d\n", ret); + goto err_usbtll_fck; + } omap->utmi_p1_fck = clk_get(dev, "utmi_p1_gfclk"); if (IS_ERR(omap->utmi_p1_fck)) { ret = PTR_ERR(omap->utmi_p1_fck); dev_err(dev, "utmi_p1_gfclk failed error:%d\n", ret); - goto err_end; + goto err_usbtll_ick; } omap->xclk60mhsp1_ck = clk_get(dev, "xclk60mhsp1_ck"); @@ -485,8 +522,22 @@ err_xclk60mhsp1_ck: err_utmi_p1_fck: clk_put(omap->utmi_p1_fck); +err_usbtll_ick: + clk_put(omap->usbtll_ick); + +err_usbtll_fck: + clk_put(omap->usbtll_fck); + +err_usbhost_fs_fck: + clk_put(omap->usbhost_fs_fck); + +err_usbhost_hs_fck: + clk_put(omap->usbhost_hs_fck); + +err_usbhost_ick: + clk_put(omap->usbhost_ick); + err_end: - pm_runtime_disable(&pdev->dev); kfree(omap); end_probe: @@ -520,7 +571,11 @@ static int __devexit usbhs_omap_remove(struct platform_device *pdev) clk_put(omap->utmi_p2_fck); clk_put(omap->xclk60mhsp1_ck); clk_put(omap->utmi_p1_fck); - pm_runtime_disable(&pdev->dev); + clk_put(omap->usbtll_ick); + clk_put(omap->usbtll_fck); + clk_put(omap->usbhost_fs_fck); + clk_put(omap->usbhost_hs_fck); + clk_put(omap->usbhost_ick); kfree(omap); return 0; @@ -640,6 +695,7 @@ static int usbhs_enable(struct device *dev) struct usbhs_omap_platform_data *pdata = &omap->platdata; unsigned long flags = 0; int ret = 0; + unsigned long timeout; unsigned reg; dev_dbg(dev, "starting TI HSUSB Controller\n"); @@ -652,7 +708,11 @@ static int usbhs_enable(struct device *dev) if (omap->count > 0) goto end_count; - pm_runtime_get_sync(dev); + clk_enable(omap->usbhost_ick); + clk_enable(omap->usbhost_hs_fck); + clk_enable(omap->usbhost_fs_fck); + clk_enable(omap->usbtll_fck); + clk_enable(omap->usbtll_ick); if (pdata->ehci_data->phy_reset) { if (gpio_is_valid(pdata->ehci_data->reset_gpio_port[0])) { @@ -676,6 +736,50 @@ static int usbhs_enable(struct device *dev) omap->usbhs_rev = usbhs_read(omap->uhh_base, OMAP_UHH_REVISION); dev_dbg(dev, "OMAP UHH_REVISION 0x%x\n", omap->usbhs_rev); + /* perform TLL soft reset, and wait until reset is complete */ + usbhs_write(omap->tll_base, OMAP_USBTLL_SYSCONFIG, + OMAP_USBTLL_SYSCONFIG_SOFTRESET); + + /* Wait for TLL reset to complete */ + timeout = jiffies + msecs_to_jiffies(1000); + while (!(usbhs_read(omap->tll_base, OMAP_USBTLL_SYSSTATUS) + & OMAP_USBTLL_SYSSTATUS_RESETDONE)) { + cpu_relax(); + + if (time_after(jiffies, timeout)) { + dev_dbg(dev, "operation timed out\n"); + ret = -EINVAL; + goto err_tll; + } + } + + dev_dbg(dev, "TLL RESET DONE\n"); + + /* (1<<3) = no idle mode only for initial debugging */ + usbhs_write(omap->tll_base, OMAP_USBTLL_SYSCONFIG, + OMAP_USBTLL_SYSCONFIG_ENAWAKEUP | + OMAP_USBTLL_SYSCONFIG_SIDLEMODE | + OMAP_USBTLL_SYSCONFIG_AUTOIDLE); + + /* Put UHH in NoIdle/NoStandby mode */ + reg = usbhs_read(omap->uhh_base, OMAP_UHH_SYSCONFIG); + if (is_omap_usbhs_rev1(omap)) { + reg |= (OMAP_UHH_SYSCONFIG_ENAWAKEUP + | OMAP_UHH_SYSCONFIG_SIDLEMODE + | OMAP_UHH_SYSCONFIG_CACTIVITY + | OMAP_UHH_SYSCONFIG_MIDLEMODE); + reg &= ~OMAP_UHH_SYSCONFIG_AUTOIDLE; + + + } else if (is_omap_usbhs_rev2(omap)) { + reg &= ~OMAP4_UHH_SYSCONFIG_IDLEMODE_CLEAR; + reg |= OMAP4_UHH_SYSCONFIG_NOIDLE; + reg &= ~OMAP4_UHH_SYSCONFIG_STDBYMODE_CLEAR; + reg |= OMAP4_UHH_SYSCONFIG_NOSTDBY; + } + + usbhs_write(omap->uhh_base, OMAP_UHH_SYSCONFIG, reg); + reg = usbhs_read(omap->uhh_base, OMAP_UHH_HOSTCONFIG); /* setup ULPI bypass and burst configurations */ reg |= (OMAP_UHH_HOSTCONFIG_INCR4_BURST_EN @@ -815,7 +919,6 @@ end_count: return 0; err_tll: - pm_runtime_put_sync(dev); spin_unlock_irqrestore(&omap->lock, flags); if (pdata->ehci_data->phy_reset) { if (gpio_is_valid(pdata->ehci_data->reset_gpio_port[0])) @@ -824,6 +927,12 @@ err_tll: if (gpio_is_valid(pdata->ehci_data->reset_gpio_port[1])) gpio_free(pdata->ehci_data->reset_gpio_port[1]); } + + clk_disable(omap->usbtll_ick); + clk_disable(omap->usbtll_fck); + clk_disable(omap->usbhost_fs_fck); + clk_disable(omap->usbhost_hs_fck); + clk_disable(omap->usbhost_ick); return ret; } @@ -896,8 +1005,6 @@ static void usbhs_disable(struct device *dev) clk_disable(omap->utmi_p1_fck); } - pm_runtime_put_sync(dev); - /* The gpio_free migh sleep; so unlock the spinlock */ spin_unlock_irqrestore(&omap->lock, flags); @@ -908,6 +1015,14 @@ static void usbhs_disable(struct device *dev) if (gpio_is_valid(pdata->ehci_data->reset_gpio_port[1])) gpio_free(pdata->ehci_data->reset_gpio_port[1]); } + + clk_disable(omap->utmi_p2_fck); + clk_disable(omap->utmi_p1_fck); + clk_disable(omap->usbtll_ick); + clk_disable(omap->usbtll_fck); + clk_disable(omap->usbhost_fs_fck); + clk_disable(omap->usbhost_hs_fck); + clk_disable(omap->usbhost_ick); return; end_disble: