| Message ID | 20180719133554.30737-1-andre.przywara@arm.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On Thu, Jul 19, 2018 at 02:35:54PM +0100, Andre Przywara wrote: > Recent Linux versions refuse to print actual virtual kernel addresses, > to not give a hint about the location of the kernel in a randomized virtual > address space. This affects the output of the sunxi MMC controller > driver, which now produces the rather uninformative line: > > [ 1.482660] sunxi-mmc 1c0f000.mmc: base:0x(____ptrval____) irq:8 > > Since the virtual base address is not really interesting in the first > place, let's just drop this value. The same applies to Linux' notion of > the interrupt number, which is independent from the GIC SPI number. > We have the physical address as part of the DT node name, which is way > more useful for debugging purposes. > To keep a success message in the driver, we print some information that > is not too obvious and that we learned while probing the device: > the maximum request size and whether it uses the new timing mode. > So the output turns into: > sunxi-mmc 1c0f000.mmc: max request size: 16384 KB, uses new timings mode > sunxi-mmc 1c11000.mmc: max request size: 2048 KB > > Signed-off-by: Andre Przywara <andre.przywara@arm.com> > --- > Changelog v1 ... v2: > - dropped output of Linux interrupt number > - added max request size and timings mode output > > drivers/mmc/host/sunxi-mmc.c | 5 ++++- > 1 file changed, 4 insertions(+), 1 deletion(-) > > diff --git a/drivers/mmc/host/sunxi-mmc.c b/drivers/mmc/host/sunxi-mmc.c > index 8e7f3e35ee3d..fbbc09d82338 100644 > --- a/drivers/mmc/host/sunxi-mmc.c > +++ b/drivers/mmc/host/sunxi-mmc.c > @@ -1407,7 +1407,10 @@ static int sunxi_mmc_probe(struct platform_device *pdev) > if (ret) > goto error_free_dma; > > - dev_info(&pdev->dev, "base:0x%p irq:%u\n", host->reg_base, host->irq); > + dev_info(&pdev->dev, "max request size: %u KB%s\n", > + mmc->max_req_size >> 10, > + host->use_new_timings ? ", uses new timings mode" : ""); I really don't know how to feel about this one. This isn't more useful to the regular user wanting to see if the driver is probed, which is what this message should be about. And this one isn't clearer or more obvious than the previous one (which was already pretty bad). I really think having some message that basically says "MMC controller initialized" or something along those lines would work better. However, I can also see value in having this printed, for developers, but maybe as dev_dbg? Maxime
Hi, On 19/07/18 15:46, Maxime Ripard wrote: > On Thu, Jul 19, 2018 at 02:35:54PM +0100, Andre Przywara wrote: >> Recent Linux versions refuse to print actual virtual kernel addresses, >> to not give a hint about the location of the kernel in a randomized virtual >> address space. This affects the output of the sunxi MMC controller >> driver, which now produces the rather uninformative line: >> >> [ 1.482660] sunxi-mmc 1c0f000.mmc: base:0x(____ptrval____) irq:8 >> >> Since the virtual base address is not really interesting in the first >> place, let's just drop this value. The same applies to Linux' notion of >> the interrupt number, which is independent from the GIC SPI number. >> We have the physical address as part of the DT node name, which is way >> more useful for debugging purposes. >> To keep a success message in the driver, we print some information that >> is not too obvious and that we learned while probing the device: >> the maximum request size and whether it uses the new timing mode. >> So the output turns into: >> sunxi-mmc 1c0f000.mmc: max request size: 16384 KB, uses new timings mode >> sunxi-mmc 1c11000.mmc: max request size: 2048 KB >> >> Signed-off-by: Andre Przywara <andre.przywara@arm.com> >> --- >> Changelog v1 ... v2: >> - dropped output of Linux interrupt number >> - added max request size and timings mode output >> >> drivers/mmc/host/sunxi-mmc.c | 5 ++++- >> 1 file changed, 4 insertions(+), 1 deletion(-) >> >> diff --git a/drivers/mmc/host/sunxi-mmc.c b/drivers/mmc/host/sunxi-mmc.c >> index 8e7f3e35ee3d..fbbc09d82338 100644 >> --- a/drivers/mmc/host/sunxi-mmc.c >> +++ b/drivers/mmc/host/sunxi-mmc.c >> @@ -1407,7 +1407,10 @@ static int sunxi_mmc_probe(struct platform_device *pdev) >> if (ret) >> goto error_free_dma; >> >> - dev_info(&pdev->dev, "base:0x%p irq:%u\n", host->reg_base, host->irq); >> + dev_info(&pdev->dev, "max request size: %u KB%s\n", >> + mmc->max_req_size >> 10, >> + host->use_new_timings ? ", uses new timings mode" : ""); > > I really don't know how to feel about this one. > > This isn't more useful to the regular user wanting to see if the > driver is probed, which is what this message should be about. > > And this one isn't clearer or more obvious than the previous one > (which was already pretty bad). I really think having some message > that basically says "MMC controller initialized" or something along > those lines would work better. I see your point, and am happy to change that. On the other hand there might be people that complain about chatty drivers, and printing a line just to says "Success!!!!!" is a bit BSP kernel like ;-). That dmesg line could as well be used to print something useful, and transport the "success" message along with that. So basically I scanned the probe function for some information that would justify an output (something non-obvious or information probed), to avoid dropping this at all (as you initially said yesterday). MMC_CAP_1_8V_DDR was another candidate, btw. > However, I can also see value in having this printed, for developers, > but maybe as dev_dbg? I think it's useful to have this in non-debug kernels as well, because this is what people tend to use. And this would allow developers to much easier debug user problems, for instance when they create board DTs: When this line doesn't appear, there might be a regulator missing, for instance. If it's there, the MMC driver is happy and we have one less thing to worry about. So why not combine the benefit of the success message and the information for developers, if we have that one line of output anyway? I think we have way more chattier drivers and way more cryptic messages in the dmesg, so a single line with some technical details does not hurt, especially if we have that already. But I don't have a strong opinion on that, so leave it up for Ulf and you to decide: keep this patch (or print some other info); replace the output with a success message or drop the line at all. Cheers, Andre.
On Thu, Jul 19, 2018 at 04:10:50PM +0100, Andre Przywara wrote: > Hi, > > On 19/07/18 15:46, Maxime Ripard wrote: > > On Thu, Jul 19, 2018 at 02:35:54PM +0100, Andre Przywara wrote: > >> Recent Linux versions refuse to print actual virtual kernel addresses, > >> to not give a hint about the location of the kernel in a randomized virtual > >> address space. This affects the output of the sunxi MMC controller > >> driver, which now produces the rather uninformative line: > >> > >> [ 1.482660] sunxi-mmc 1c0f000.mmc: base:0x(____ptrval____) irq:8 > >> > >> Since the virtual base address is not really interesting in the first > >> place, let's just drop this value. The same applies to Linux' notion of > >> the interrupt number, which is independent from the GIC SPI number. > >> We have the physical address as part of the DT node name, which is way > >> more useful for debugging purposes. > >> To keep a success message in the driver, we print some information that > >> is not too obvious and that we learned while probing the device: > >> the maximum request size and whether it uses the new timing mode. > >> So the output turns into: > >> sunxi-mmc 1c0f000.mmc: max request size: 16384 KB, uses new timings mode > >> sunxi-mmc 1c11000.mmc: max request size: 2048 KB > >> > >> Signed-off-by: Andre Przywara <andre.przywara@arm.com> > >> --- > >> Changelog v1 ... v2: > >> - dropped output of Linux interrupt number > >> - added max request size and timings mode output > >> > >> drivers/mmc/host/sunxi-mmc.c | 5 ++++- > >> 1 file changed, 4 insertions(+), 1 deletion(-) > >> > >> diff --git a/drivers/mmc/host/sunxi-mmc.c b/drivers/mmc/host/sunxi-mmc.c > >> index 8e7f3e35ee3d..fbbc09d82338 100644 > >> --- a/drivers/mmc/host/sunxi-mmc.c > >> +++ b/drivers/mmc/host/sunxi-mmc.c > >> @@ -1407,7 +1407,10 @@ static int sunxi_mmc_probe(struct platform_device *pdev) > >> if (ret) > >> goto error_free_dma; > >> > >> - dev_info(&pdev->dev, "base:0x%p irq:%u\n", host->reg_base, host->irq); > >> + dev_info(&pdev->dev, "max request size: %u KB%s\n", > >> + mmc->max_req_size >> 10, > >> + host->use_new_timings ? ", uses new timings mode" : ""); > > > > I really don't know how to feel about this one. > > > > This isn't more useful to the regular user wanting to see if the > > driver is probed, which is what this message should be about. > > > > And this one isn't clearer or more obvious than the previous one > > (which was already pretty bad). I really think having some message > > that basically says "MMC controller initialized" or something along > > those lines would work better. > > I see your point, and am happy to change that. > On the other hand there might be people that complain about chatty > drivers, and printing a line just to says "Success!!!!!" is a bit BSP > kernel like ;-). Well, we're currently printing copyright notices, so I guess we're worse than the BSPs :) > That dmesg line could as well be used to print something useful, and > transport the "success" message along with that. So basically I > scanned the probe function for some information that would justify > an output (something non-obvious or information probed), to avoid > dropping this at all (as you initially said yesterday). > MMC_CAP_1_8V_DDR was another candidate, btw. > > > However, I can also see value in having this printed, for developers, > > but maybe as dev_dbg? > > I think it's useful to have this in non-debug kernels as well, because > this is what people tend to use. And this would allow developers to much > easier debug user problems, for instance when they create board DTs: > When this line doesn't appear, there might be a regulator missing, for > instance. If it's there, the MMC driver is happy and we have one less > thing to worry about. > > So why not combine the benefit of the success message and the > information for developers, if we have that one line of output anyway? > I think we have way more chattier drivers and way more cryptic messages > in the dmesg, so a single line with some technical details does not > hurt, especially if we have that already. > > But I don't have a strong opinion on that, so leave it up for Ulf and > you to decide: keep this patch (or print some other info); replace the > output with a success message or drop the line at all. How about both then? Something like "initialized with %d request size and new timings\n" ? It makes it obvious enough for someone that doesn't really has an idea, while outputting the proper informations you wanted. Maxime
Hi, On 20/07/18 15:17, Maxime Ripard wrote: > On Thu, Jul 19, 2018 at 04:10:50PM +0100, Andre Przywara wrote: >> Hi, >> >> On 19/07/18 15:46, Maxime Ripard wrote: >>> On Thu, Jul 19, 2018 at 02:35:54PM +0100, Andre Przywara wrote: >>>> Recent Linux versions refuse to print actual virtual kernel addresses, >>>> to not give a hint about the location of the kernel in a randomized virtual >>>> address space. This affects the output of the sunxi MMC controller >>>> driver, which now produces the rather uninformative line: >>>> >>>> [ 1.482660] sunxi-mmc 1c0f000.mmc: base:0x(____ptrval____) irq:8 >>>> >>>> Since the virtual base address is not really interesting in the first >>>> place, let's just drop this value. The same applies to Linux' notion of >>>> the interrupt number, which is independent from the GIC SPI number. >>>> We have the physical address as part of the DT node name, which is way >>>> more useful for debugging purposes. >>>> To keep a success message in the driver, we print some information that >>>> is not too obvious and that we learned while probing the device: >>>> the maximum request size and whether it uses the new timing mode. >>>> So the output turns into: >>>> sunxi-mmc 1c0f000.mmc: max request size: 16384 KB, uses new timings mode >>>> sunxi-mmc 1c11000.mmc: max request size: 2048 KB >>>> >>>> Signed-off-by: Andre Przywara <andre.przywara@arm.com> >>>> --- >>>> Changelog v1 ... v2: >>>> - dropped output of Linux interrupt number >>>> - added max request size and timings mode output >>>> >>>> drivers/mmc/host/sunxi-mmc.c | 5 ++++- >>>> 1 file changed, 4 insertions(+), 1 deletion(-) >>>> >>>> diff --git a/drivers/mmc/host/sunxi-mmc.c b/drivers/mmc/host/sunxi-mmc.c >>>> index 8e7f3e35ee3d..fbbc09d82338 100644 >>>> --- a/drivers/mmc/host/sunxi-mmc.c >>>> +++ b/drivers/mmc/host/sunxi-mmc.c >>>> @@ -1407,7 +1407,10 @@ static int sunxi_mmc_probe(struct platform_device *pdev) >>>> if (ret) >>>> goto error_free_dma; >>>> >>>> - dev_info(&pdev->dev, "base:0x%p irq:%u\n", host->reg_base, host->irq); >>>> + dev_info(&pdev->dev, "max request size: %u KB%s\n", >>>> + mmc->max_req_size >> 10, >>>> + host->use_new_timings ? ", uses new timings mode" : ""); >>> >>> I really don't know how to feel about this one. >>> >>> This isn't more useful to the regular user wanting to see if the >>> driver is probed, which is what this message should be about. >>> >>> And this one isn't clearer or more obvious than the previous one >>> (which was already pretty bad). I really think having some message >>> that basically says "MMC controller initialized" or something along >>> those lines would work better. >> >> I see your point, and am happy to change that. >> On the other hand there might be people that complain about chatty >> drivers, and printing a line just to says "Success!!!!!" is a bit BSP >> kernel like ;-). > > Well, we're currently printing copyright notices, so I guess we're > worse than the BSPs :) > >> That dmesg line could as well be used to print something useful, and >> transport the "success" message along with that. So basically I >> scanned the probe function for some information that would justify >> an output (something non-obvious or information probed), to avoid >> dropping this at all (as you initially said yesterday). >> MMC_CAP_1_8V_DDR was another candidate, btw. >> >>> However, I can also see value in having this printed, for developers, >>> but maybe as dev_dbg? >> >> I think it's useful to have this in non-debug kernels as well, because >> this is what people tend to use. And this would allow developers to much >> easier debug user problems, for instance when they create board DTs: >> When this line doesn't appear, there might be a regulator missing, for >> instance. If it's there, the MMC driver is happy and we have one less >> thing to worry about. >> >> So why not combine the benefit of the success message and the >> information for developers, if we have that one line of output anyway? >> I think we have way more chattier drivers and way more cryptic messages >> in the dmesg, so a single line with some technical details does not >> hurt, especially if we have that already. >> >> But I don't have a strong opinion on that, so leave it up for Ulf and >> you to decide: keep this patch (or print some other info); replace the >> output with a success message or drop the line at all. > > How about both then? > > Something like "initialized with %d request size and new timings\n" ? > > It makes it obvious enough for someone that doesn't really has an > idea, while outputting the proper informations you wanted. Excellent idea. Changed it in this direction. Thanks, Andre.
diff --git a/drivers/mmc/host/sunxi-mmc.c b/drivers/mmc/host/sunxi-mmc.c index 8e7f3e35ee3d..fbbc09d82338 100644 --- a/drivers/mmc/host/sunxi-mmc.c +++ b/drivers/mmc/host/sunxi-mmc.c @@ -1407,7 +1407,10 @@ static int sunxi_mmc_probe(struct platform_device *pdev) if (ret) goto error_free_dma; - dev_info(&pdev->dev, "base:0x%p irq:%u\n", host->reg_base, host->irq); + dev_info(&pdev->dev, "max request size: %u KB%s\n", + mmc->max_req_size >> 10, + host->use_new_timings ? ", uses new timings mode" : ""); + return 0; error_free_dma:
Recent Linux versions refuse to print actual virtual kernel addresses, to not give a hint about the location of the kernel in a randomized virtual address space. This affects the output of the sunxi MMC controller driver, which now produces the rather uninformative line: [ 1.482660] sunxi-mmc 1c0f000.mmc: base:0x(____ptrval____) irq:8 Since the virtual base address is not really interesting in the first place, let's just drop this value. The same applies to Linux' notion of the interrupt number, which is independent from the GIC SPI number. We have the physical address as part of the DT node name, which is way more useful for debugging purposes. To keep a success message in the driver, we print some information that is not too obvious and that we learned while probing the device: the maximum request size and whether it uses the new timing mode. So the output turns into: sunxi-mmc 1c0f000.mmc: max request size: 16384 KB, uses new timings mode sunxi-mmc 1c11000.mmc: max request size: 2048 KB Signed-off-by: Andre Przywara <andre.przywara@arm.com> --- Changelog v1 ... v2: - dropped output of Linux interrupt number - added max request size and timings mode output drivers/mmc/host/sunxi-mmc.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-)