| Message ID | 20230623144100.34196-1-james.quinlan@broadcom.com (mailing list archive) |
|---|---|
| Headers | show |
| Series | PCI: brcmstb: Configure appropriate HW CLKREQ# mode | expand |
Hi Jim, Jim Quinlan <james.quinlan@broadcom.com> (2023-06-23): > v6 -- No code has been changed. > -- Changed commit subject and comment in "#PERST" commit (Bjorn, Cyril) > -- Changed sign-off and author email address for all commits. > This was due to a change in Broadcom's upstreaming policy. I've just run some more tests to be on the safe side, and I can confirm everything is still looking good with the updated series and the updated base commit. Test setup: ----------- - using a $CM with the 20230111 EEPROM - on the same CM4 IO Board - with a $PCIE board (PCIe to multiple USB ports) - and the same Samsung USB flash drive. where $CM is one of: - CM4 Lite Rev 1.0 - CM4 8/32 Rev 1.0 - CM4 4/32 Rev 1.1 and $PCIE is one of: - SupaHub PCE6U1C-R02, VER 006 - SupaHub PCE6U1C-R02, VER 006S Results: -------- 1. With an unpatched kernel, I'm getting the dreaded Serror for all $CM/$PCIE combinations. That's reproducible with: - the 6.1.y kernel shipped in Debian 12; - a locally-built v6.4-rc7-194-g8a28a0b6f1a1d kernel. 2. With a patched kernel (v6.4-rc7-194-g8a28a0b6f1a1d + this series), for all $CM/$PCIE combinations, I'm getting a system that boots, sees the flash drive, and gives decent read performance on the USB flash drive (200+ MB/s on the CM4 Lite, 220+ MB/s on the non-Lite versions). In passing, since that looks like it could be merged finally: I suppose it's fair to say this series adds support for hardware that wasn't working before, which means it's not a candidate for inclusion via stable@ (even if it gets rid of a nasty failure to boot depending on what hardware is plugged in at that time)? In other words, downstream distributions should be expected to either adjust their build systems to pick some future Linux release or consider backporting this series on their own, to each base Linux version they support? Thanks again for all the help figuring this out. Cheers,
On Fri, Jun 23, 2023 at 10:40:53AM -0400, Jim Quinlan wrote: > v6 -- No code has been changed. > -- Changed commit subject and comment in "#PERST" commit (Bjorn, Cyril) > -- Changed sign-off and author email address for all commits. > This was due to a change in Broadcom's upstreaming policy. > > v5 -- Remove DT property "brcm,completion-timeout-us" from > "DT bindings" commit. Although this error may be reported > as a completion timeout, its cause was traced to an > internal bus timeout which may occur even when there is > no PCIe access being processed. We set a timeout of four > seconds only if we are operating in "L1SS CLKREQ#" mode. > -- Correct CEM 2.0 reference provided by HW engineer, > s/3.2.5.2.5/3.2.5.2.2/ (Bjorn) > -- Add newline to dev_info() string (Stefan) > -- Change variable rval to unsigned (Stefan) > -- s/implementaion/implementation/ (Bjorn) > -- s/superpowersave/powersupersave/ (Bjorn) > -- Slightly modify message on "PERST#" commit. > -- Rebase to torvalds master > > v4 -- New commit that asserts PERST# for 2711/RPi SOCs at PCIe RC > driver probe() time. This is done in Raspian Linux and its > absence may be the cause of a failing test case. > -- New commit that removes stale comment. > > v3 -- Rewrote commit msgs and comments refering panics if L1SS > is enabled/disabled; the code snippet that unadvertises L1SS > eliminates the panic scenario. (Bjorn) > -- Add reference for "400ns of CLKREQ# assertion" blurb (Bjorn) > -- Put binding names in DT commit Subject (Bjorn) > -- Add a verb to a commit's subject line (Bjorn) > -- s/accomodat(\w+)/accommodat$1/g (Bjorn) > -- Rewrote commit msgs and comments refering panics if L1SS > is enabled/disabled; the code snippet that unadvertises L1SS > eliminates the panic scenario. (Bjorn) > > v2 -- Changed binding property 'brcm,completion-timeout-msec' to > 'brcm,completion-timeout-us'. (StefanW for standard suffix). > -- Warn when clamping timeout value, and include clamped > region in message. Also add min and max in YAML. (StefanW) > -- Qualify description of "brcm,completion-timeout-us" so that > it refers to PCIe transactions. (StefanW) > -- Remvove mention of Linux specifics in binding description. (StefanW) > -- s/clkreq#/CLKREQ#/g (Bjorn) > -- Refactor completion-timeout-us code to compare max and min to > value given by the property (as opposed to the computed value). > > v1 -- The current driver assumes the downstream devices can > provide CLKREQ# for ASPM. These commits accomodate devices > w/ or w/o clkreq# and also handle L1SS-capable devices. > > -- The Raspian Linux folks have already been using a PCIe RC > property "brcm,enable-l1ss". These commits use the same > property, in a backward-compatible manner, and the implementaion > adds more detail and also automatically identifies devices w/o > a clkreq# signal, i.e. most devices plugged into an RPi CM4 > IO board. > > > Jim Quinlan (5): > dt-bindings: PCI: brcmstb: Add brcm,enable-l1ss property > PCI: brcmstb: Configure HW CLKREQ# mode appropriate for downstream > device I am not merging the first two patches since the discussion thread is still open and I'd like to understand better what can/should be done, sorry. > PCI: brcmstb: Set higher value for internal bus timeout > PCI: brcmstb: Assert PERST# on BCM2711 > PCI: brcmstb: Remove stale comment Is it OK to apply these three on their own ? Overall it would be great to avoid mixing patches with different end goals in a single series. Thanks, Lorenzo > .../bindings/pci/brcm,stb-pcie.yaml | 9 ++ > drivers/pci/controller/pcie-brcmstb.c | 91 ++++++++++++++++--- > 2 files changed, 89 insertions(+), 11 deletions(-) > > > base-commit: 8a28a0b6f1a1dcbf5a834600a9acfbe2ba51e5eb > -- > 2.17.1 >
On Mon, Aug 21, 2023 at 4:35 AM Lorenzo Pieralisi <lpieralisi@kernel.org> wrote: > > On Fri, Jun 23, 2023 at 10:40:53AM -0400, Jim Quinlan wrote: > > v6 -- No code has been changed. > > -- Changed commit subject and comment in "#PERST" commit (Bjorn, Cyril) > > -- Changed sign-off and author email address for all commits. > > This was due to a change in Broadcom's upstreaming policy. > > > > v5 -- Remove DT property "brcm,completion-timeout-us" from > > "DT bindings" commit. Although this error may be reported > > as a completion timeout, its cause was traced to an > > internal bus timeout which may occur even when there is > > no PCIe access being processed. We set a timeout of four > > seconds only if we are operating in "L1SS CLKREQ#" mode. > > -- Correct CEM 2.0 reference provided by HW engineer, > > s/3.2.5.2.5/3.2.5.2.2/ (Bjorn) > > -- Add newline to dev_info() string (Stefan) > > -- Change variable rval to unsigned (Stefan) > > -- s/implementaion/implementation/ (Bjorn) > > -- s/superpowersave/powersupersave/ (Bjorn) > > -- Slightly modify message on "PERST#" commit. > > -- Rebase to torvalds master > > > > v4 -- New commit that asserts PERST# for 2711/RPi SOCs at PCIe RC > > driver probe() time. This is done in Raspian Linux and its > > absence may be the cause of a failing test case. > > -- New commit that removes stale comment. > > > > v3 -- Rewrote commit msgs and comments refering panics if L1SS > > is enabled/disabled; the code snippet that unadvertises L1SS > > eliminates the panic scenario. (Bjorn) > > -- Add reference for "400ns of CLKREQ# assertion" blurb (Bjorn) > > -- Put binding names in DT commit Subject (Bjorn) > > -- Add a verb to a commit's subject line (Bjorn) > > -- s/accomodat(\w+)/accommodat$1/g (Bjorn) > > -- Rewrote commit msgs and comments refering panics if L1SS > > is enabled/disabled; the code snippet that unadvertises L1SS > > eliminates the panic scenario. (Bjorn) > > > > v2 -- Changed binding property 'brcm,completion-timeout-msec' to > > 'brcm,completion-timeout-us'. (StefanW for standard suffix). > > -- Warn when clamping timeout value, and include clamped > > region in message. Also add min and max in YAML. (StefanW) > > -- Qualify description of "brcm,completion-timeout-us" so that > > it refers to PCIe transactions. (StefanW) > > -- Remvove mention of Linux specifics in binding description. (StefanW) > > -- s/clkreq#/CLKREQ#/g (Bjorn) > > -- Refactor completion-timeout-us code to compare max and min to > > value given by the property (as opposed to the computed value). > > > > v1 -- The current driver assumes the downstream devices can > > provide CLKREQ# for ASPM. These commits accomodate devices > > w/ or w/o clkreq# and also handle L1SS-capable devices. > > > > -- The Raspian Linux folks have already been using a PCIe RC > > property "brcm,enable-l1ss". These commits use the same > > property, in a backward-compatible manner, and the implementaion > > adds more detail and also automatically identifies devices w/o > > a clkreq# signal, i.e. most devices plugged into an RPi CM4 > > IO board. > > > > > > Jim Quinlan (5): > > dt-bindings: PCI: brcmstb: Add brcm,enable-l1ss property > > PCI: brcmstb: Configure HW CLKREQ# mode appropriate for downstream > > device > > I am not merging the first two patches since the discussion thread > is still open and I'd like to understand better what can/should be > done, sorry. Hello Lorenzo, This patch-set has been stable for months, V5 was out early May and the V6 changes did not involve code. I'm a little surprised that you are voicing concern at this stage. The previous discussions covered all aspects of these commits AFAICT. Please review them and the commit messages and let me know what issues you do not understand or any topics that were not considered. Are you concerned about the Broadcom STB/CM community or the RPi community? For the former, I have direct communication w/ our customers and none of them are even close to using upstream (they may backport my commits). For the latter, I have tested these commits on the official RPi4 and CM4 IO platforms, and Cyril has also put in an admiral amount of testing. Note that I have on my desk a CM4 IO board w/ a conventional PCIe device, and it does not boot upstream master Linux until these patches are applied. Further, Raspian OS has already introduced the "brcm,enable-l1ss" property but did not upstream it, and my commits are backwards compatible with this. > > > PCI: brcmstb: Set higher value for internal bus timeout > > PCI: brcmstb: Assert PERST# on BCM2711 > > PCI: brcmstb: Remove stale comment > > Is it OK to apply these three on their own ? Overall it would be > great to avoid mixing patches with different end goals in a single > series. Well, they are related for one customer who wants to use L1SS power savings AND require a long period for the internal timeout. But, yes, these commits are fine to apply independently. Regards, Jim Quinlan Broadcom STB > > Thanks, > Lorenzo > > > .../bindings/pci/brcm,stb-pcie.yaml | 9 ++ > > drivers/pci/controller/pcie-brcmstb.c | 91 ++++++++++++++++--- > > 2 files changed, 89 insertions(+), 11 deletions(-) > > > > > > base-commit: 8a28a0b6f1a1dcbf5a834600a9acfbe2ba51e5eb > > -- > > 2.17.1 > > > >
On Mon, Aug 21, 2023 at 08:15:02AM -0400, Jim Quinlan wrote: > On Mon, Aug 21, 2023 at 4:35 AM Lorenzo Pieralisi <lpieralisi@kernel.org> wrote: > > > > On Fri, Jun 23, 2023 at 10:40:53AM -0400, Jim Quinlan wrote: > > > v6 -- No code has been changed. > > > -- Changed commit subject and comment in "#PERST" commit (Bjorn, Cyril) > > > -- Changed sign-off and author email address for all commits. > > > This was due to a change in Broadcom's upstreaming policy. > > > > > > v5 -- Remove DT property "brcm,completion-timeout-us" from > > > "DT bindings" commit. Although this error may be reported > > > as a completion timeout, its cause was traced to an > > > internal bus timeout which may occur even when there is > > > no PCIe access being processed. We set a timeout of four > > > seconds only if we are operating in "L1SS CLKREQ#" mode. > > > -- Correct CEM 2.0 reference provided by HW engineer, > > > s/3.2.5.2.5/3.2.5.2.2/ (Bjorn) > > > -- Add newline to dev_info() string (Stefan) > > > -- Change variable rval to unsigned (Stefan) > > > -- s/implementaion/implementation/ (Bjorn) > > > -- s/superpowersave/powersupersave/ (Bjorn) > > > -- Slightly modify message on "PERST#" commit. > > > -- Rebase to torvalds master > > > > > > v4 -- New commit that asserts PERST# for 2711/RPi SOCs at PCIe RC > > > driver probe() time. This is done in Raspian Linux and its > > > absence may be the cause of a failing test case. > > > -- New commit that removes stale comment. > > > > > > v3 -- Rewrote commit msgs and comments refering panics if L1SS > > > is enabled/disabled; the code snippet that unadvertises L1SS > > > eliminates the panic scenario. (Bjorn) > > > -- Add reference for "400ns of CLKREQ# assertion" blurb (Bjorn) > > > -- Put binding names in DT commit Subject (Bjorn) > > > -- Add a verb to a commit's subject line (Bjorn) > > > -- s/accomodat(\w+)/accommodat$1/g (Bjorn) > > > -- Rewrote commit msgs and comments refering panics if L1SS > > > is enabled/disabled; the code snippet that unadvertises L1SS > > > eliminates the panic scenario. (Bjorn) > > > > > > v2 -- Changed binding property 'brcm,completion-timeout-msec' to > > > 'brcm,completion-timeout-us'. (StefanW for standard suffix). > > > -- Warn when clamping timeout value, and include clamped > > > region in message. Also add min and max in YAML. (StefanW) > > > -- Qualify description of "brcm,completion-timeout-us" so that > > > it refers to PCIe transactions. (StefanW) > > > -- Remvove mention of Linux specifics in binding description. (StefanW) > > > -- s/clkreq#/CLKREQ#/g (Bjorn) > > > -- Refactor completion-timeout-us code to compare max and min to > > > value given by the property (as opposed to the computed value). > > > > > > v1 -- The current driver assumes the downstream devices can > > > provide CLKREQ# for ASPM. These commits accomodate devices > > > w/ or w/o clkreq# and also handle L1SS-capable devices. > > > > > > -- The Raspian Linux folks have already been using a PCIe RC > > > property "brcm,enable-l1ss". These commits use the same > > > property, in a backward-compatible manner, and the implementaion > > > adds more detail and also automatically identifies devices w/o > > > a clkreq# signal, i.e. most devices plugged into an RPi CM4 > > > IO board. > > > > > > > > > Jim Quinlan (5): > > > dt-bindings: PCI: brcmstb: Add brcm,enable-l1ss property > > > PCI: brcmstb: Configure HW CLKREQ# mode appropriate for downstream > > > device > > > > I am not merging the first two patches since the discussion thread > > is still open and I'd like to understand better what can/should be > > done, sorry. > > Hello Lorenzo, > > This patch-set has been stable for months, V5 was out early May and > the V6 changes > did not involve code. I'm a little surprised that you are voicing > concern at this stage. > > The previous discussions covered all aspects of these commits AFAICT. > Please review > them and the commit messages and let me know what issues you do not understand > or any topics that were not considered. I disagree with the reasoning behind "brcm,enable-l1ss" property usage instead of a command line option - at least I would like to get a comment from DT maintainers about it. I think Bjorn made the point consistently and I also think he is right. I would like to get Rob's opinion on this. I know he acked the DT bindings (I have a comment on those too) but regardless, it is clearly a property used for what is a command line configuration parameter, no two ways about it. Thanks, Lorenzo > > Are you concerned about the Broadcom STB/CM community or the RPi community? > For the former, I have direct communication w/ our customers and none of them > are even close to using upstream (they may backport my commits). For > the latter, I have > tested these commits on the official RPi4 and CM4 IO platforms, and > Cyril has also put in > an admiral amount of testing. > > Note that I have on my desk a CM4 IO board w/ a conventional PCIe > device, and it does not boot > upstream master Linux until these patches are applied. > > Further, Raspian OS has already introduced the "brcm,enable-l1ss" > property but did not upstream it, and > my commits are backwards compatible with this. > > > > > > PCI: brcmstb: Set higher value for internal bus timeout > > > PCI: brcmstb: Assert PERST# on BCM2711 > > > PCI: brcmstb: Remove stale comment > > > > Is it OK to apply these three on their own ? Overall it would be > > great to avoid mixing patches with different end goals in a single > > series. > > Well, they are related for one customer who wants to use L1SS power > savings AND require > a long period for the internal timeout. But, yes, these commits are > fine to apply > independently. > > Regards, > Jim Quinlan > Broadcom STB > > > > > Thanks, > > Lorenzo > > > > > .../bindings/pci/brcm,stb-pcie.yaml | 9 ++ > > > drivers/pci/controller/pcie-brcmstb.c | 91 ++++++++++++++++--- > > > 2 files changed, 89 insertions(+), 11 deletions(-) > > > > > > > > > base-commit: 8a28a0b6f1a1dcbf5a834600a9acfbe2ba51e5eb > > > -- > > > 2.17.1 > > > > > > >
On Fri, 23 Jun 2023 10:40:53 -0400, Jim Quinlan wrote: > v6 -- No code has been changed. > -- Changed commit subject and comment in "#PERST" commit (Bjorn, Cyril) > -- Changed sign-off and author email address for all commits. > This was due to a change in Broadcom's upstreaming policy. > > v5 -- Remove DT property "brcm,completion-timeout-us" from > "DT bindings" commit. Although this error may be reported > as a completion timeout, its cause was traced to an > internal bus timeout which may occur even when there is > no PCIe access being processed. We set a timeout of four > seconds only if we are operating in "L1SS CLKREQ#" mode. > -- Correct CEM 2.0 reference provided by HW engineer, > s/3.2.5.2.5/3.2.5.2.2/ (Bjorn) > -- Add newline to dev_info() string (Stefan) > -- Change variable rval to unsigned (Stefan) > -- s/implementaion/implementation/ (Bjorn) > -- s/superpowersave/powersupersave/ (Bjorn) > -- Slightly modify message on "PERST#" commit. > -- Rebase to torvalds master > > [...] Applied to controller/brcmstb, thanks! [4/5] PCI: brcmstb: Assert PERST# on BCM2711 https://git.kernel.org/pci/pci/c/8eb8c2735306 [5/5] PCI: brcmstb: Remove stale comment https://git.kernel.org/pci/pci/c/6dac1507a654 Thanks, Lorenzo
v6 -- No code has been changed. -- Changed commit subject and comment in "#PERST" commit (Bjorn, Cyril) -- Changed sign-off and author email address for all commits. This was due to a change in Broadcom's upstreaming policy. v5 -- Remove DT property "brcm,completion-timeout-us" from "DT bindings" commit. Although this error may be reported as a completion timeout, its cause was traced to an internal bus timeout which may occur even when there is no PCIe access being processed. We set a timeout of four seconds only if we are operating in "L1SS CLKREQ#" mode. -- Correct CEM 2.0 reference provided by HW engineer, s/3.2.5.2.5/3.2.5.2.2/ (Bjorn) -- Add newline to dev_info() string (Stefan) -- Change variable rval to unsigned (Stefan) -- s/implementaion/implementation/ (Bjorn) -- s/superpowersave/powersupersave/ (Bjorn) -- Slightly modify message on "PERST#" commit. -- Rebase to torvalds master v4 -- New commit that asserts PERST# for 2711/RPi SOCs at PCIe RC driver probe() time. This is done in Raspian Linux and its absence may be the cause of a failing test case. -- New commit that removes stale comment. v3 -- Rewrote commit msgs and comments refering panics if L1SS is enabled/disabled; the code snippet that unadvertises L1SS eliminates the panic scenario. (Bjorn) -- Add reference for "400ns of CLKREQ# assertion" blurb (Bjorn) -- Put binding names in DT commit Subject (Bjorn) -- Add a verb to a commit's subject line (Bjorn) -- s/accomodat(\w+)/accommodat$1/g (Bjorn) -- Rewrote commit msgs and comments refering panics if L1SS is enabled/disabled; the code snippet that unadvertises L1SS eliminates the panic scenario. (Bjorn) v2 -- Changed binding property 'brcm,completion-timeout-msec' to 'brcm,completion-timeout-us'. (StefanW for standard suffix). -- Warn when clamping timeout value, and include clamped region in message. Also add min and max in YAML. (StefanW) -- Qualify description of "brcm,completion-timeout-us" so that it refers to PCIe transactions. (StefanW) -- Remvove mention of Linux specifics in binding description. (StefanW) -- s/clkreq#/CLKREQ#/g (Bjorn) -- Refactor completion-timeout-us code to compare max and min to value given by the property (as opposed to the computed value). v1 -- The current driver assumes the downstream devices can provide CLKREQ# for ASPM. These commits accomodate devices w/ or w/o clkreq# and also handle L1SS-capable devices. -- The Raspian Linux folks have already been using a PCIe RC property "brcm,enable-l1ss". These commits use the same property, in a backward-compatible manner, and the implementaion adds more detail and also automatically identifies devices w/o a clkreq# signal, i.e. most devices plugged into an RPi CM4 IO board. Jim Quinlan (5): dt-bindings: PCI: brcmstb: Add brcm,enable-l1ss property PCI: brcmstb: Configure HW CLKREQ# mode appropriate for downstream device PCI: brcmstb: Set higher value for internal bus timeout PCI: brcmstb: Assert PERST# on BCM2711 PCI: brcmstb: Remove stale comment .../bindings/pci/brcm,stb-pcie.yaml | 9 ++ drivers/pci/controller/pcie-brcmstb.c | 91 ++++++++++++++++--- 2 files changed, 89 insertions(+), 11 deletions(-) base-commit: 8a28a0b6f1a1dcbf5a834600a9acfbe2ba51e5eb