RE: [PATCHv5 6/6] PCI: layerscape: Add power management support

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



Hi Krzysztof,

Thanks a lot for your comments!

> -----Original Message-----
> From: Krzysztof Wilczyński [mailto:kw@xxxxxxxxx]
> Sent: 2021年11月12日 8:37
> To: Z.Q. Hou <zhiqiang.hou@xxxxxxx>
> Cc: linux-pci@xxxxxxxxxxxxxxx; devicetree@xxxxxxxxxxxxxxx;
> linux-kernel@xxxxxxxxxxxxxxx; linux-arm-kernel@xxxxxxxxxxxxxxxxxxx;
> lorenzo.pieralisi@xxxxxxx; robh+dt@xxxxxxxxxx; bhelgaas@xxxxxxxxxx;
> shawnguo@xxxxxxxxxx; Leo Li <leoyang.li@xxxxxxx>;
> gustavo.pimentel@xxxxxxxxxxxx; M.H. Lian <minghuan.lian@xxxxxxx>;
> Mingkai Hu <mingkai.hu@xxxxxxx>; Roy Zang <roy.zang@xxxxxxx>
> Subject: Re: [PATCHv5 6/6] PCI: layerscape: Add power management support
> 
> Hi,
> 
> [...]
> > +/* PF Message Command Register */
> > +#define LS_PCIE_PF_MCR		0x2c
> > +#define PF_MCR_PTOMR		BIT(0)
> > +#define PF_MCR_EXL2S		BIT(1)
> > +
> > +/* LS1021A PEXn PM Write Control Register */
> > +#define SCFG_PEXPMWRCR(idx)	(0x5c + (idx) * 0x64)
> > +#define PMXMTTURNOFF		BIT(31)
> > +#define SCFG_PEXSFTRSTCR	0x190
> > +#define PEXSR(idx)		BIT(idx)
> > +
> > +/* LS1043A PEX PME control register */
> > +#define SCFG_PEXPMECR		0x144
> > +#define PEXPME(idx)		BIT(31 - (idx) * 4)
> > +
> > +/* LS1043A PEX LUT debug register */
> > +#define LS_PCIE_LDBG	0x7fc
> > +#define LDBG_SR		BIT(30)
> > +#define LDBG_WE		BIT(31)
> 
> A small nitpick: a consistent capitalisation of "control" and "debug", and
> "register" in the comments above.

Good suggestion, will make them consistent next version.

> 
> [...]
> > +static void ls_pcie_lut_writel(struct ls_pcie *pcie, u32 off, u32
> > +val) {
> > +	if (pcie->big_endian)
> > +		return iowrite32be(val, pcie->lut_base + off);
> > +
> > +	return iowrite32(val, pcie->lut_base + off);
> > +
> > +}
> 
> Surplus newline above after the return statement.
> 
> [...]
> > +static void ls_pcie_pf_writel(struct ls_pcie *pcie, u32 off, u32 val)
> > +{
> > +	if (pcie->big_endian)
> > +		return iowrite32be(val, pcie->pf_base + off);
> > +
> > +	return iowrite32(val, pcie->pf_base + off);
> > +
> > +}
> 
> Surplus newline above after the return statement.
> 

Will remove these lines next version.

> [...]
> > +static void ls_pcie_send_turnoff_msg(struct ls_pcie *pcie) {
> > +	u32 val;
> > +	int ret;
> > +
> > +	val = ls_pcie_pf_readl(pcie, LS_PCIE_PF_MCR);
> > +	val |= PF_MCR_PTOMR;
> > +	ls_pcie_pf_writel(pcie, LS_PCIE_PF_MCR, val);
> > +
> > +	ret = readx_poll_timeout(ls_pcie_pf_readl_addr, LS_PCIE_PF_MCR,
> > +				 val, !(val & PF_MCR_PTOMR), 100, 10000);
> > +	if (ret)
> > +		dev_info(pcie->pci->dev, "poll turn off message timeout\n"); }
> 
> Would this dev_info() be more of a warning or an error?  A timeout is
> potentially a problem, correct?
>

An error message is better here, will change next version.
 
> [...]
> > +static void ls1021a_pcie_send_turnoff_msg(struct ls_pcie *pcie) {
> > +	u32 val;
> > +
> > +	if (!pcie->scfg) {
> > +		dev_dbg(pcie->pci->dev, "SYSCFG is NULL\n");
> > +		return;
> > +	}
> > +
> > +	/* Send Turn_off message */
> > +	regmap_read(pcie->scfg, SCFG_PEXPMWRCR(pcie->index), &val);
> > +	val |= PMXMTTURNOFF;
> > +	regmap_write(pcie->scfg, SCFG_PEXPMWRCR(pcie->index), val);
> > +
> > +	mdelay(10);
> 
> We often, customary, document why a particular mdelay() is needed.  You also
> did this in other part of the code, so perhaps adding a note here (and
> everywhere else) would be nice for keeping the consistency.
> 

Will add next version.

> [...]
> > +static void ls_pcie_exit_from_l2(struct ls_pcie *pcie) {
> > +	u32 val;
> > +	int ret;
> > +
> > +	val = ls_pcie_pf_readl(pcie, LS_PCIE_PF_MCR);
> > +	val |= PF_MCR_EXL2S;
> > +	ls_pcie_pf_writel(pcie, LS_PCIE_PF_MCR, val);
> > +
> > +	ret = readx_poll_timeout(ls_pcie_pf_readl_addr, LS_PCIE_PF_MCR,
> > +				 val, !(val & PF_MCR_EXL2S), 100, 10000);
> > +	if (ret)
> > +		dev_info(pcie->pci->dev, "poll exit L2 state timeout\n"); }
> 
> Similarly to the question above: is this timeout something more severe and
> would warrant a warning or an error here instead?
> 

Agree.

> [...]
> > +static void ls1021a_pcie_exit_from_l2(struct ls_pcie *pcie) {
> > +	u32 val;
> > +
> > +	regmap_read(pcie->scfg, SCFG_PEXSFTRSTCR, &val);
> > +	val |= PEXSR(pcie->index);
> > +	regmap_write(pcie->scfg, SCFG_PEXSFTRSTCR, val);
> > +
> > +	regmap_read(pcie->scfg, SCFG_PEXSFTRSTCR, &val);
> > +	val &= ~PEXSR(pcie->index);
> > +	regmap_write(pcie->scfg, SCFG_PEXSFTRSTCR, val);
> > +
> > +	mdelay(1);
> 
> Aside of documenting this mdelay() here, if possible, would 1 be enough?
> Everywhere else you seem to use 10 consistently.
> 

It's enough and didn't encounter a fail in thousands regressions.

> > +
> > +	ls_pcie_retrain_link(pcie);
> > +}
> > +static void ls1043a_pcie_exit_from_l2(struct ls_pcie *pcie)
> 
> Missing newline above to separate code blocks.
> 

Will add next version.

> > +{
> > +	u32 val;
> > +
> > +	val = ls_pcie_lut_readl(pcie, LS_PCIE_LDBG);
> > +	val |= LDBG_WE;
> > +	ls_pcie_lut_writel(pcie, LS_PCIE_LDBG, val);
> > +
> > +	val = ls_pcie_lut_readl(pcie, LS_PCIE_LDBG);
> > +	val |= LDBG_SR;
> > +	ls_pcie_lut_writel(pcie, LS_PCIE_LDBG, val);
> > +
> > +	val = ls_pcie_lut_readl(pcie, LS_PCIE_LDBG);
> > +	val &= ~LDBG_SR;
> > +	ls_pcie_lut_writel(pcie, LS_PCIE_LDBG, val);
> > +
> > +	val = ls_pcie_lut_readl(pcie, LS_PCIE_LDBG);
> > +	val &= ~LDBG_WE;
> > +	ls_pcie_lut_writel(pcie, LS_PCIE_LDBG, val);
> > +
> > +	mdelay(1);
> 
> See comment above.
> 
> [...]
> > +static int ls1021a_pcie_pm_init(struct ls_pcie *pcie) {
> > +	struct device *dev = pcie->pci->dev;
> > +	u32 index[2];
> > +	int ret;
> > +
> > +	pcie->scfg = syscon_regmap_lookup_by_phandle(dev->of_node,
> > +						     "fsl,pcie-scfg");
> > +	if (IS_ERR(pcie->scfg)) {
> > +		ret = PTR_ERR(pcie->scfg);
> > +		dev_err(dev, "No syscfg phandle specified\n");
> > +		pcie->scfg = NULL;
> > +		return ret;
> > +	}
> > +
> > +	ret = of_property_read_u32_array(dev->of_node, "fsl,pcie-scfg",
> > +					 index, 2);
> > +	if (ret) {
> > +		pcie->scfg = NULL;
> > +		return ret;
> > +	}
> > +
> > +	pcie->index = index[1];
> > +
> > +	return 0;
> > +}
> 
> Just an idea: what about using goto for error handling?
> 
> (...)
> 	if (IS_ERR(pcie->scfg)) {
> 		ret = PTR_ERR(pcie->scfg);
> 		dev_err(dev, "No syscfg phandle specified\n");
> 		goto error;
> 	}
> 
> 	ret = of_property_read_u32_array(dev->of_node, "fsl,pcie-scfg",
> 					 index, 2);
> 	if (ret)
> 		goto error;
> 
> 	pcie->index = index[1];
> 
> 	return 0;
> 
> error:
> 	pcie->scfg = NULL;
> 	return ret;
> }
> 
> Not necessarily better or worse compared with your version, so it would be
> more of a matter of personal preference here.

Thanks for the good suggestion!

> 
> > +static int ls_pcie_suspend_noirq(struct device *dev) {
> > +	struct ls_pcie *pcie = dev_get_drvdata(dev);
> > +	struct dw_pcie *pci = pcie->pci;
> > +	u32 val;
> > +	int ret;
> > +
> > +	if (!ls_pcie_pm_check(pcie))
> > +		return 0;
> > +
> > +	pcie->drvdata->pm_ops->send_turn_off_message(pcie);
> > +
> > +	/* 10ms timeout to check L2 ready */
> > +	ret = readl_poll_timeout(pci->dbi_base + PCIE_PORT_DEBUG0,
> > +				 val, LS_PCIE_IS_L2(val), 100, 10000);
> > +	if (ret) {
> > +		dev_err(dev, "PCIe link enter L2 timeout! ltssm = 0x%x\n", val);
> > +		return ret;
> > +	}
> 
> The error message above could be improve to be more like an error stating that
> something failed and such, as currently it looks like a debug message, unless it
> was intended as such.

Exactly, will improve.

> 
> [...]
> > +static int ls_pcie_resume_noirq(struct device *dev) {
> > +	struct ls_pcie *pcie = dev_get_drvdata(dev);
> > +	struct dw_pcie *pci = pcie->pci;
> > +	int ret;
> > +
> > +	if (!ls_pcie_pm_check(pcie))
> > +		return 0;
> > +
> > +	ls_pcie_set_dstate(pcie, 0x0);
> > +
> > +	pcie->drvdata->pm_ops->exit_from_l2(pcie);
> > +
> > +	dw_pcie_setup_rc(&pci->pp);
> > +
> > +	/* delay 10 ms to access EP */
> > +	mdelay(10);
> > +
> > +	ret = ls_pcie_host_init(&pci->pp);
> > +	if (ret) {
> > +		dev_err(dev, "ls_pcie_host_init failed! ret = 0x%x\n", ret);
> > +		return ret;
> > +	}
> 
> A small nitpick: error messages that are directed at end users should have a little
> more context than just the function name.

Will remove the return value check, currently this func always succeed.

Thanks,
Zhiqiang

> 
> 	Krzysztof




[Index of Archives]     [DMA Engine]     [Linux Coverity]     [Linux USB]     [Video for Linux]     [Linux Audio Users]     [Yosemite News]     [Linux Kernel]     [Linux SCSI]     [Greybus]

  Powered by Linux