| Message ID | 20201022154506.17639-1-joakim.tjernlund@infinera.com |
|---|---|
| State | New |
| Headers | show |
| Series | mtd: cfi_cmdset_0002: Use status register where possible | expand |
Hi Joakim, Please Cc the MTD maintainers, not only the list (get_maintainers.pl). Joakim Tjernlund <joakim.tjernlund@infinera.com> wrote on Thu, 22 Oct 2020 17:45:06 +0200: > Commit "mtd: cfi_cmdset_0002: Add support for polling status register" > added support for polling the status rather than using DQ polling. > However, status register is used only when DQ polling is missing. > Lets use status register when available as it is superior to DQ polling. > I will let vignesh comment about the content (looks fine by me) but you will need a Fixes tag here. > Signed-off-by: Joakim Tjernlund <joakim.tjernlund@infinera.com> > Cc: stable@vger.kernel.org > --- > drivers/mtd/chips/cfi_cmdset_0002.c | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/drivers/mtd/chips/cfi_cmdset_0002.c b/drivers/mtd/chips/cfi_cmdset_0002.c > index a1f3e1031c3d..ee9b322e63bb 100644 > --- a/drivers/mtd/chips/cfi_cmdset_0002.c > +++ b/drivers/mtd/chips/cfi_cmdset_0002.c > @@ -117,7 +117,7 @@ static struct mtd_chip_driver cfi_amdstd_chipdrv = { > static int cfi_use_status_reg(struct cfi_private *cfi) > { > struct cfi_pri_amdstd *extp = cfi->cmdset_priv; > - u8 poll_mask = CFI_POLL_STATUS_REG | CFI_POLL_DQ; > + u8 poll_mask = CFI_POLL_STATUS_REG; > > return extp->MinorVersion >= '5' && > (extp->SoftwareFeatures & poll_mask) == CFI_POLL_STATUS_REG; Thanks, Miquèl
On Fri, 2020-10-30 at 18:47 +0100, Miquel Raynal wrote: > > Hi Joakim, Hi Miquel > > Please Cc the MTD maintainers, not only the list (get_maintainers.pl). I figure all maintainers are on the list ? > > Joakim Tjernlund <joakim.tjernlund@infinera.com> wrote on Thu, 22 Oct > 2020 17:45:06 +0200: > > > Commit "mtd: cfi_cmdset_0002: Add support for polling status register" > > added support for polling the status rather than using DQ polling. > > However, status register is used only when DQ polling is missing. > > Lets use status register when available as it is superior to DQ polling. > > > > I will let vignesh comment about the content (looks fine by me) but you will > need a Fixes tag here. This is not a Fixes situation, no bug just a hw enabling thing. Also, I would like to see the Status patches be backported to 4.19 as well. Jocke
Hi Joakim, Joakim Tjernlund <Joakim.Tjernlund@infinera.com> wrote on Fri, 30 Oct 2020 18:39:35 +0000: > On Fri, 2020-10-30 at 18:47 +0100, Miquel Raynal wrote: > > > > Hi Joakim, > > Hi Miquel > > > > > Please Cc the MTD maintainers, not only the list (get_maintainers.pl). > > I figure all maintainers are on the list ? I personally don't look at the list very often. I expect patches to be directed to me (in the current case, Vignesh) when I am concerned. > > > > > Joakim Tjernlund <joakim.tjernlund@infinera.com> wrote on Thu, 22 Oct > > 2020 17:45:06 +0200: > > > > > Commit "mtd: cfi_cmdset_0002: Add support for polling status register" > > > added support for polling the status rather than using DQ polling. > > > However, status register is used only when DQ polling is missing. > > > Lets use status register when available as it is superior to DQ polling. > > > > > > > I will let vignesh comment about the content (looks fine by me) but you will > > need a Fixes tag here. > > This is not a Fixes situation, no bug just a hw enabling thing. > Also, I would like to see the Status patches be backported to 4.19 as well. Backporting features is IMHO not relevant. I guess stable kernel only take fixes... Thanks, Miquèl
On Fri, 2020-10-30 at 19:52 +0100, Miquel Raynal wrote: > CAUTION: This email originated from outside of the organization. Do not click links or open attachments unless you recognize the sender and know the content is safe. > > > Hi Joakim, > > Joakim Tjernlund <Joakim.Tjernlund@infinera.com> wrote on Fri, 30 Oct > 2020 18:39:35 +0000: > > > On Fri, 2020-10-30 at 18:47 +0100, Miquel Raynal wrote: > > > > > > Hi Joakim, > > > > Hi Miquel > > > > > > > > Please Cc the MTD maintainers, not only the list (get_maintainers.pl). > > > > I figure all maintainers are on the list ? > > I personally don't look at the list very often. I expect patches to be > directed to me (in the current case, Vignesh) when I am concerned. Added Vignesh > > > > > > > > > Joakim Tjernlund <joakim.tjernlund@infinera.com> wrote on Thu, 22 Oct > > > 2020 17:45:06 +0200: > > > > > > > Commit "mtd: cfi_cmdset_0002: Add support for polling status register" > > > > added support for polling the status rather than using DQ polling. > > > > However, status register is used only when DQ polling is missing. > > > > Lets use status register when available as it is superior to DQ polling. > > > > > > > > > > I will let vignesh comment about the content (looks fine by me) but you will > > > need a Fixes tag here. > > > > This is not a Fixes situation, no bug just a hw enabling thing. > > Also, I would like to see the Status patches be backported to 4.19 as well. > > Backporting features is IMHO not relevant. I guess stable kernel only > take fixes... This is not a feature really and the 5.4 stable did get them, I ask 4.19 get them too.
Hi Joakim On 10/31/20 4:56 PM, Joakim Tjernlund wrote: > On Fri, 2020-10-30 at 19:52 +0100, Miquel Raynal wrote: >> CAUTION: This email originated from outside of the organization. Do not click links or open attachments unless you recognize the sender and know the content is safe. >> >> >> Hi Joakim, >> >> Joakim Tjernlund <Joakim.Tjernlund@infinera.com> wrote on Fri, 30 Oct >> 2020 18:39:35 +0000: >> >>> On Fri, 2020-10-30 at 18:47 +0100, Miquel Raynal wrote: >>>> >>>> Hi Joakim, >>> >>> Hi Miquel >>> >>>> >>>> Please Cc the MTD maintainers, not only the list (get_maintainers.pl). >>> >>> I figure all maintainers are on the list ? >> >> I personally don't look at the list very often. I expect patches to be >> directed to me (in the current case, Vignesh) when I am concerned. > > Added Vignesh > As Miquel suggested, I look at patches on mailing list at a lower priority than patches that are CC'd to me. >> >>> >>>> >>>> Joakim Tjernlund <joakim.tjernlund@infinera.com> wrote on Thu, 22 Oct >>>> 2020 17:45:06 +0200: >>>> >>>>> Commit "mtd: cfi_cmdset_0002: Add support for polling status register" >>>>> added support for polling the status rather than using DQ polling. >>>>> However, status register is used only when DQ polling is missing. >>>>> Lets use status register when available as it is superior to DQ polling. >>>>> >>>> >>>> I will let vignesh comment about the content (looks fine by me) but you will >>>> need a Fixes tag here. >>> >>> This is not a Fixes situation, no bug just a hw enabling thing. >>> Also, I would like to see the Status patches be backported to 4.19 as well. >> >> Backporting features is IMHO not relevant. I guess stable kernel only >> take fixes... > > This is not a feature really and the 5.4 stable did get them, I ask 4.19 get them too. > commit 4844ef80305d ("mtd: cfi_cmdset_0002: Add support for polling status register") was added in 5.3 and therefore is part of 5.4. But note that this is a "new feature" and therefore won't be backported to kernels older than 5.3. Similarly, this patch (when accepted) is not a candidate for stable kernel backports because the intention of enabling polling status register for Write Completion is to enable flashes that "don't" support DQ polling at all (mainly HyperFlash). Enabling this for all flashes that support the feature is not a bug fix IMO. Also, there isn't enough testing to prove that feature works for all CFI NOR flashes on all platforms and therefore would be risky to be backported to stable kernels. Regards Vignehs
On Wed, 2020-11-04 at 11:12 +0530, Vignesh Raghavendra wrote: > Hi Joakim > > On 10/31/20 4:56 PM, Joakim Tjernlund wrote: > > On Fri, 2020-10-30 at 19:52 +0100, Miquel Raynal wrote: > > > CAUTION: This email originated from outside of the organization. Do not click links or open attachments unless you recognize the sender and know the content is safe. > > > > > > > > > Hi Joakim, > > > > > > Joakim Tjernlund <Joakim.Tjernlund@infinera.com> wrote on Fri, 30 Oct > > > 2020 18:39:35 +0000: > > > > > > > On Fri, 2020-10-30 at 18:47 +0100, Miquel Raynal wrote: > > > > > > > > > > Hi Joakim, > > > > > > > > Hi Miquel > > > > > > > > > > > > > > Please Cc the MTD maintainers, not only the list (get_maintainers.pl). > > > > > > > > I figure all maintainers are on the list ? > > > > > > I personally don't look at the list very often. I expect patches to be > > > directed to me (in the current case, Vignesh) when I am concerned. > > > > Added Vignesh > > > > As Miquel suggested, I look at patches on mailing list at a lower > priority than patches that are CC'd to me. > > > > > > > > > > > > > > > > > > > Joakim Tjernlund <joakim.tjernlund@infinera.com> wrote on Thu, 22 Oct > > > > > 2020 17:45:06 +0200: > > > > > > > > > > > Commit "mtd: cfi_cmdset_0002: Add support for polling status register" > > > > > > added support for polling the status rather than using DQ polling. > > > > > > However, status register is used only when DQ polling is missing. > > > > > > Lets use status register when available as it is superior to DQ polling. > > > > > > > > > > > > > > > > I will let vignesh comment about the content (looks fine by me) but you will > > > > > need a Fixes tag here. > > > > > > > > This is not a Fixes situation, no bug just a hw enabling thing. > > > > Also, I would like to see the Status patches be backported to 4.19 as well. > > > > > > Backporting features is IMHO not relevant. I guess stable kernel only > > > take fixes... > > > > This is not a feature really and the 5.4 stable did get them, I ask 4.19 get them too. > > > > commit 4844ef80305d ("mtd: cfi_cmdset_0002: Add support for polling > status register") was added in 5.3 and therefore is part of 5.4. But > note that this is a "new feature" and therefore won't be backported to > kernels older than 5.3. Oh, my memory is off then, sorry. > Similarly, this patch (when accepted) is not a candidate for stable > kernel backports because the intention of enabling polling status > register for Write Completion is to enable flashes that "don't" support > DQ polling at all (mainly HyperFlash). > Enabling this for all flashes that support the feature is not a bug fix > IMO. Also, there isn't enough testing to prove that feature works for > all CFI NOR flashes on all platforms and therefore would be risky to be > backported to stable kernels. This is 2 things, 1. making it possible to use Hyper flash (HW enablement) 2. improving the flash driver to function more precise(getting an accurate error rather than a TMO) for most AMD flashes. 1. happens on a regular basis in stable. 2. can be discussed for stable but should at least go into master now that status has been in there for a while Jocke
Hi Joakim, On 10/22/20 9:15 PM, Joakim Tjernlund wrote: > Commit "mtd: cfi_cmdset_0002: Add support for polling status register" Standard way to refer to a commit is: Commit 4844ef80305d ("mtd: cfi_cmdset_0002: Add support for polling status register") > added support for polling the status rather than using DQ polling. > However, status register is used only when DQ polling is missing. > Lets use status register when available as it is superior to DQ polling. > > Signed-off-by: Joakim Tjernlund <joakim.tjernlund@infinera.com> > Cc: stable@vger.kernel.org > --- > drivers/mtd/chips/cfi_cmdset_0002.c | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/drivers/mtd/chips/cfi_cmdset_0002.c b/drivers/mtd/chips/cfi_cmdset_0002.c > index a1f3e1031c3d..ee9b322e63bb 100644 > --- a/drivers/mtd/chips/cfi_cmdset_0002.c > +++ b/drivers/mtd/chips/cfi_cmdset_0002.c > @@ -117,7 +117,7 @@ static struct mtd_chip_driver cfi_amdstd_chipdrv = { > static int cfi_use_status_reg(struct cfi_private *cfi) > { > struct cfi_pri_amdstd *extp = cfi->cmdset_priv; > - u8 poll_mask = CFI_POLL_STATUS_REG | CFI_POLL_DQ; > + u8 poll_mask = CFI_POLL_STATUS_REG; This local variable now looks pointless and can be dropped.. So, > return extp->MinorVersion >= '5' && > (extp->SoftwareFeatures & poll_mask) == CFI_POLL_STATUS_REG; > return extp->MinorVersion >= '5' && (extp->SoftwareFeatures & CFI_POLL_STATUS_REG); Regards Vignesh
On 11/4/20 3:23 PM, Joakim Tjernlund wrote: > On Wed, 2020-11-04 at 11:12 +0530, Vignesh Raghavendra wrote: >> Hi Joakim >> >> On 10/31/20 4:56 PM, Joakim Tjernlund wrote: >>> On Fri, 2020-10-30 at 19:52 +0100, Miquel Raynal wrote: >>>> CAUTION: This email originated from outside of the organization. Do not click links or open attachments unless you recognize the sender and know the content is safe. >>>> >>>> [...] >> commit 4844ef80305d ("mtd: cfi_cmdset_0002: Add support for polling >> status register") was added in 5.3 and therefore is part of 5.4. But >> note that this is a "new feature" and therefore won't be backported to >> kernels older than 5.3. > > Oh, my memory is off then, sorry. > >> Similarly, this patch (when accepted) is not a candidate for stable >> kernel backports because the intention of enabling polling status >> register for Write Completion is to enable flashes that "don't" support >> DQ polling at all (mainly HyperFlash). >> Enabling this for all flashes that support the feature is not a bug fix >> IMO. Also, there isn't enough testing to prove that feature works for >> all CFI NOR flashes on all platforms and therefore would be risky to be >> backported to stable kernels. > > This is 2 things, > 1. making it possible to use Hyper flash (HW enablement) > 2. improving the flash driver to function more precise(getting an accurate error rather than a TMO) for most AMD flashes. > > 1. happens on a regular basis in stable. > 2. can be discussed for stable but should at least go into master now that status has been in there for a while > Yes, but my objection is that there has not been enough testing to prove that its safe to backport to stable kernels and there is nothing that's broken in older kernels which this patch fixes. Regards Vignesh
On Thu, 2020-11-05 at 21:34 +0530, Vignesh Raghavendra wrote: > > On 11/4/20 3:23 PM, Joakim Tjernlund wrote: > > On Wed, 2020-11-04 at 11:12 +0530, Vignesh Raghavendra wrote: > > > Hi Joakim > > > > > > On 10/31/20 4:56 PM, Joakim Tjernlund wrote: > > > > On Fri, 2020-10-30 at 19:52 +0100, Miquel Raynal wrote: > > > > > CAUTION: This email originated from outside of the organization. Do not click links or open attachments unless you recognize the sender and know the content is safe. > > > > > > > > > > > [...] > > > commit 4844ef80305d ("mtd: cfi_cmdset_0002: Add support for polling > > > status register") was added in 5.3 and therefore is part of 5.4. But > > > note that this is a "new feature" and therefore won't be backported to > > > kernels older than 5.3. > > > > Oh, my memory is off then, sorry. > > > > > Similarly, this patch (when accepted) is not a candidate for stable > > > kernel backports because the intention of enabling polling status > > > register for Write Completion is to enable flashes that "don't" support > > > DQ polling at all (mainly HyperFlash). > > > Enabling this for all flashes that support the feature is not a bug fix > > > IMO. Also, there isn't enough testing to prove that feature works for > > > all CFI NOR flashes on all platforms and therefore would be risky to be > > > backported to stable kernels. > > > > This is 2 things, > > 1. making it possible to use Hyper flash (HW enablement) > > 2. improving the flash driver to function more precise(getting an accurate error rather than a TMO) for most AMD flashes. > > > > 1. happens on a regular basis in stable. > > 2. can be discussed for stable but should at least go into master now that status has been in there for a while > > > > Yes, but my objection is that there has not been enough testing to prove > that its safe to backport to stable kernels and there is nothing that's > broken in older kernels which this patch fixes. Semi broken though. I stumbled upon some erase failures that just resulted in a super long TMO error rather than report a failure directly(the page was locked). There I no way to improve upon error handling without your status updates and this is the main reason I want your patches backported to 4.19 as well. My patch can be dropped for stable, giving it time to mature in master though. Jocke
diff --git a/drivers/mtd/chips/cfi_cmdset_0002.c b/drivers/mtd/chips/cfi_cmdset_0002.c index a1f3e1031c3d..ee9b322e63bb 100644 --- a/drivers/mtd/chips/cfi_cmdset_0002.c +++ b/drivers/mtd/chips/cfi_cmdset_0002.c @@ -117,7 +117,7 @@ static struct mtd_chip_driver cfi_amdstd_chipdrv = { static int cfi_use_status_reg(struct cfi_private *cfi) { struct cfi_pri_amdstd *extp = cfi->cmdset_priv; - u8 poll_mask = CFI_POLL_STATUS_REG | CFI_POLL_DQ; + u8 poll_mask = CFI_POLL_STATUS_REG; return extp->MinorVersion >= '5' && (extp->SoftwareFeatures & poll_mask) == CFI_POLL_STATUS_REG;
Commit "mtd: cfi_cmdset_0002: Add support for polling status register" added support for polling the status rather than using DQ polling. However, status register is used only when DQ polling is missing. Lets use status register when available as it is superior to DQ polling. Signed-off-by: Joakim Tjernlund <joakim.tjernlund@infinera.com> Cc: stable@vger.kernel.org --- drivers/mtd/chips/cfi_cmdset_0002.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)