Message ID | 20181204101021.22817-4-stefan@agner.ch |
---|---|
State | Accepted |
Commit | 52c2c97e7c5b3ba326bae53a7940e27878efd405 |
Delegated to: | Stefano Babic |
Headers | show |
Series | ddr: vybrid: various fixes | expand |
Hi Stefan, > From: Stefan Agner <stefan.agner@toradex.com> > > The CR80 register has multiple interrupt bits, the code is supposed > to check bit 8 but instead uses a logical and. In most cases this > probably did not affect real operations since at that stage typically > none of the other bits are set. It can exit the loop when any error bit is set (but anyway with broken DDR initialization we hang latter anyway). > > Signed-off-by: Stefan Agner <stefan.agner@toradex.com> > Acked-by: Marcel Ziswiler <marcel.ziswiler@toradex.com> > --- > > arch/arm/include/asm/arch-vf610/imx-regs.h | 3 ++- > arch/arm/mach-imx/ddrmc-vf610.c | 3 ++- > 2 files changed, 4 insertions(+), 2 deletions(-) > > diff --git a/arch/arm/include/asm/arch-vf610/imx-regs.h > b/arch/arm/include/asm/arch-vf610/imx-regs.h index > b7374bfb8f..f71fbf4e73 100644 --- > a/arch/arm/include/asm/arch-vf610/imx-regs.h +++ > b/arch/arm/include/asm/arch-vf610/imx-regs.h @@ -200,7 +200,8 @@ > #define DDRMC_CR78_Q_FULLNESS(v) (((v) & 0x7) > << 24) #define DDRMC_CR78_BUR_ON_FLY_BIT(v) > ((v) & 0xf) #define DDRMC_CR79_CTLUPD_AREF(v) > (((v) & 0x1) << 24) -#define > DDRMC_CR82_INT_MASK 0x10000000 > +#define DDRMC_CR80_MC_INIT_COMPLETE (1 << 8) > +#define DDRMC_CR82_INT_MASK (1 << 28) When I was working on my code - the checkpatch.pl was complaining about (1 << x) code. It was recommended to use BIT(x) instead. > #define DDRMC_CR87_ODT_WR_MAPCS0(v) ((v) << > 24) #define DDRMC_CR87_ODT_RD_MAPCS0(v) ((v) > << 16) #define DDRMC_CR88_TODTL_CMD(v) > (((v) & 0x1f) << 16) diff --git a/arch/arm/mach-imx/ddrmc-vf610.c > b/arch/arm/mach-imx/ddrmc-vf610.c index 3d7da1c25e..d121a53898 100644 > --- a/arch/arm/mach-imx/ddrmc-vf610.c +++ > b/arch/arm/mach-imx/ddrmc-vf610.c @@ -231,6 +231,7 @@ void > ddrmc_ctrl_init_ddr3(struct ddr3_jedec_timings const *timings, /* all > inits done, start the DDR controller */ > writel(DDRMC_CR00_DRAM_CLASS_DDR3 | DDRMC_CR00_START, &ddrmr->cr[0]); > - while (!(readl(&ddrmr->cr[80]) && 0x100)) > + while (!(readl(&ddrmr->cr[80]) & > DDRMC_CR80_MC_INIT_COMPLETE)) udelay(10); > + writel(DDRMC_CR80_MC_INIT_COMPLETE, &ddrmr->cr[81]); > } Best regards, Lukasz Majewski -- DENX Software Engineering GmbH, Managing Director: Wolfgang Denk HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany Phone: (+49)-8142-66989-59 Fax: (+49)-8142-66989-80 Email: lukma@denx.de
On 04.12.2018 11:59, Lukasz Majewski wrote: > Hi Stefan, > >> From: Stefan Agner <stefan.agner@toradex.com> >> >> The CR80 register has multiple interrupt bits, the code is supposed >> to check bit 8 but instead uses a logical and. In most cases this >> probably did not affect real operations since at that stage typically >> none of the other bits are set. > > It can exit the loop when any error bit is set (but anyway with broken > DDR initialization we hang latter anyway). > >> >> Signed-off-by: Stefan Agner <stefan.agner@toradex.com> >> Acked-by: Marcel Ziswiler <marcel.ziswiler@toradex.com> >> --- >> >> arch/arm/include/asm/arch-vf610/imx-regs.h | 3 ++- >> arch/arm/mach-imx/ddrmc-vf610.c | 3 ++- >> 2 files changed, 4 insertions(+), 2 deletions(-) >> >> diff --git a/arch/arm/include/asm/arch-vf610/imx-regs.h >> b/arch/arm/include/asm/arch-vf610/imx-regs.h index >> b7374bfb8f..f71fbf4e73 100644 --- >> a/arch/arm/include/asm/arch-vf610/imx-regs.h +++ >> b/arch/arm/include/asm/arch-vf610/imx-regs.h @@ -200,7 +200,8 @@ >> #define DDRMC_CR78_Q_FULLNESS(v) (((v) & 0x7) >> << 24) #define DDRMC_CR78_BUR_ON_FLY_BIT(v) >> ((v) & 0xf) #define DDRMC_CR79_CTLUPD_AREF(v) >> (((v) & 0x1) << 24) -#define >> DDRMC_CR82_INT_MASK 0x10000000 >> +#define DDRMC_CR80_MC_INIT_COMPLETE (1 << 8) >> +#define DDRMC_CR82_INT_MASK (1 << 28) > > When I was working on my code - the checkpatch.pl was complaining about > (1 << x) code. It was recommended to use BIT(x) instead. > Yeah I noticed that too, but arch/arm/include/asm/arch-vf610/imx-regs.h does use bit shifting so far. I rated consistency higher than the new recommendation :-) -- Stefan >> #define DDRMC_CR87_ODT_WR_MAPCS0(v) ((v) << >> 24) #define DDRMC_CR87_ODT_RD_MAPCS0(v) ((v) >> << 16) #define DDRMC_CR88_TODTL_CMD(v) >> (((v) & 0x1f) << 16) diff --git a/arch/arm/mach-imx/ddrmc-vf610.c >> b/arch/arm/mach-imx/ddrmc-vf610.c index 3d7da1c25e..d121a53898 100644 >> --- a/arch/arm/mach-imx/ddrmc-vf610.c +++ >> b/arch/arm/mach-imx/ddrmc-vf610.c @@ -231,6 +231,7 @@ void >> ddrmc_ctrl_init_ddr3(struct ddr3_jedec_timings const *timings, /* all >> inits done, start the DDR controller */ >> writel(DDRMC_CR00_DRAM_CLASS_DDR3 | DDRMC_CR00_START, &ddrmr->cr[0]); >> - while (!(readl(&ddrmr->cr[80]) && 0x100)) >> + while (!(readl(&ddrmr->cr[80]) & >> DDRMC_CR80_MC_INIT_COMPLETE)) udelay(10); >> + writel(DDRMC_CR80_MC_INIT_COMPLETE, &ddrmr->cr[81]); >> } > > > > > Best regards, > > Lukasz Majewski > > -- > > DENX Software Engineering GmbH, Managing Director: Wolfgang Denk > HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany > Phone: (+49)-8142-66989-59 Fax: (+49)-8142-66989-80 Email: lukma@denx.de
Lukasz, Stefano, On 04.12.2018 12:20, Stefan Agner wrote: > On 04.12.2018 11:59, Lukasz Majewski wrote: >> Hi Stefan, >> >>> From: Stefan Agner <stefan.agner@toradex.com> >>> >>> The CR80 register has multiple interrupt bits, the code is supposed >>> to check bit 8 but instead uses a logical and. In most cases this >>> probably did not affect real operations since at that stage typically >>> none of the other bits are set. >> >> It can exit the loop when any error bit is set (but anyway with broken >> DDR initialization we hang latter anyway). >> >>> >>> Signed-off-by: Stefan Agner <stefan.agner@toradex.com> >>> Acked-by: Marcel Ziswiler <marcel.ziswiler@toradex.com> >>> --- >>> >>> arch/arm/include/asm/arch-vf610/imx-regs.h | 3 ++- >>> arch/arm/mach-imx/ddrmc-vf610.c | 3 ++- >>> 2 files changed, 4 insertions(+), 2 deletions(-) >>> >>> diff --git a/arch/arm/include/asm/arch-vf610/imx-regs.h >>> b/arch/arm/include/asm/arch-vf610/imx-regs.h index >>> b7374bfb8f..f71fbf4e73 100644 --- >>> a/arch/arm/include/asm/arch-vf610/imx-regs.h +++ >>> b/arch/arm/include/asm/arch-vf610/imx-regs.h @@ -200,7 +200,8 @@ >>> #define DDRMC_CR78_Q_FULLNESS(v) (((v) & 0x7) >>> << 24) #define DDRMC_CR78_BUR_ON_FLY_BIT(v) >>> ((v) & 0xf) #define DDRMC_CR79_CTLUPD_AREF(v) >>> (((v) & 0x1) << 24) -#define >>> DDRMC_CR82_INT_MASK 0x10000000 >>> +#define DDRMC_CR80_MC_INIT_COMPLETE (1 << 8) >>> +#define DDRMC_CR82_INT_MASK (1 << 28) >> >> When I was working on my code - the checkpatch.pl was complaining about >> (1 << x) code. It was recommended to use BIT(x) instead. >> > > Yeah I noticed that too, but arch/arm/include/asm/arch-vf610/imx-regs.h > does use bit shifting so far. I rated consistency higher than the new > recommendation :-) Do you want me to change the whole file to BIT here? I'd rather prefer to keep it for now. But I think we otherwise are ok with this patch set right? I guess technically patch 4 would not be your realm Stefano, but maybe you can also take it through your tree? -- Stefan > > -- > Stefan > >>> #define DDRMC_CR87_ODT_WR_MAPCS0(v) ((v) << >>> 24) #define DDRMC_CR87_ODT_RD_MAPCS0(v) ((v) >>> << 16) #define DDRMC_CR88_TODTL_CMD(v) >>> (((v) & 0x1f) << 16) diff --git a/arch/arm/mach-imx/ddrmc-vf610.c >>> b/arch/arm/mach-imx/ddrmc-vf610.c index 3d7da1c25e..d121a53898 100644 >>> --- a/arch/arm/mach-imx/ddrmc-vf610.c +++ >>> b/arch/arm/mach-imx/ddrmc-vf610.c @@ -231,6 +231,7 @@ void >>> ddrmc_ctrl_init_ddr3(struct ddr3_jedec_timings const *timings, /* all >>> inits done, start the DDR controller */ >>> writel(DDRMC_CR00_DRAM_CLASS_DDR3 | DDRMC_CR00_START, &ddrmr->cr[0]); >>> - while (!(readl(&ddrmr->cr[80]) && 0x100)) >>> + while (!(readl(&ddrmr->cr[80]) & >>> DDRMC_CR80_MC_INIT_COMPLETE)) udelay(10); >>> + writel(DDRMC_CR80_MC_INIT_COMPLETE, &ddrmr->cr[81]); >>> } >> >> >> >> >> Best regards, >> >> Lukasz Majewski >> >> -- >> >> DENX Software Engineering GmbH, Managing Director: Wolfgang Denk >> HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany >> Phone: (+49)-8142-66989-59 Fax: (+49)-8142-66989-80 Email: lukma@denx.de
diff --git a/arch/arm/include/asm/arch-vf610/imx-regs.h b/arch/arm/include/asm/arch-vf610/imx-regs.h index b7374bfb8f..f71fbf4e73 100644 --- a/arch/arm/include/asm/arch-vf610/imx-regs.h +++ b/arch/arm/include/asm/arch-vf610/imx-regs.h @@ -200,7 +200,8 @@ #define DDRMC_CR78_Q_FULLNESS(v) (((v) & 0x7) << 24) #define DDRMC_CR78_BUR_ON_FLY_BIT(v) ((v) & 0xf) #define DDRMC_CR79_CTLUPD_AREF(v) (((v) & 0x1) << 24) -#define DDRMC_CR82_INT_MASK 0x10000000 +#define DDRMC_CR80_MC_INIT_COMPLETE (1 << 8) +#define DDRMC_CR82_INT_MASK (1 << 28) #define DDRMC_CR87_ODT_WR_MAPCS0(v) ((v) << 24) #define DDRMC_CR87_ODT_RD_MAPCS0(v) ((v) << 16) #define DDRMC_CR88_TODTL_CMD(v) (((v) & 0x1f) << 16) diff --git a/arch/arm/mach-imx/ddrmc-vf610.c b/arch/arm/mach-imx/ddrmc-vf610.c index 3d7da1c25e..d121a53898 100644 --- a/arch/arm/mach-imx/ddrmc-vf610.c +++ b/arch/arm/mach-imx/ddrmc-vf610.c @@ -231,6 +231,7 @@ void ddrmc_ctrl_init_ddr3(struct ddr3_jedec_timings const *timings, /* all inits done, start the DDR controller */ writel(DDRMC_CR00_DRAM_CLASS_DDR3 | DDRMC_CR00_START, &ddrmr->cr[0]); - while (!(readl(&ddrmr->cr[80]) && 0x100)) + while (!(readl(&ddrmr->cr[80]) & DDRMC_CR80_MC_INIT_COMPLETE)) udelay(10); + writel(DDRMC_CR80_MC_INIT_COMPLETE, &ddrmr->cr[81]); }