diff mbox series

[v2] EDAC: synopsys: Fix the wrong call of pinf->col parameter

Message ID 1582856183-5007-1-git-send-email-sherry.sun@nxp.com (mailing list archive)
State New, archived
Headers show
Series [v2] EDAC: synopsys: Fix the wrong call of pinf->col parameter | expand

Commit Message

Sherry Sun Feb. 28, 2020, 2:16 a.m. UTC
Since ZynqMP platform call zynqmp_get_error_info() function to get ce/ue
information. In this function, pinf->col parameter is not used, this
parameter is only used by Zynq platforme in zynq_get_error_info(). So
here pinf->col should not be called and printed for ZynqMP, need remove
it.

Fixes: b500b4a029d57 ("EDAC, synopsys: Add ECC support for ZynqMP DDR controller")
Signed-off-by: Sherry Sun <sherry.sun@nxp.com>
---
This patch is based on [PATCH V2] EDAC: synopsys: Fix back to back snprintf() messages for CE/UE
Changes in V2:
- Separated this patch from the original patchset.

---
 drivers/edac/synopsys_edac.c | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

Comments

Manish Narani March 9, 2020, 8:44 a.m. UTC | #1
Hi,

> -----Original Message-----
> From: Sherry Sun <sherry.sun@nxp.com>
> Sent: Friday, February 28, 2020 7:46 AM
> To: bp@alien8.de; mchehab@kernel.org; tony.luck@intel.com;
> james.morse@arm.com; rrichter@marvell.com; Michal Simek
> <michals@xilinx.com>; Manish Narani <MNARANI@xilinx.com>
> Cc: linux-edac@vger.kernel.org; linux-kernel@vger.kernel.org; linux-
> imx@nxp.com; frank.li@nxp.com
> Subject: [PATCH v2] EDAC: synopsys: Fix the wrong call of pinf->col parameter
> 
> Since ZynqMP platform call zynqmp_get_error_info() function to get ce/ue
> information. In this function, pinf->col parameter is not used, this
> parameter is only used by Zynq platforme in zynq_get_error_info(). So
> here pinf->col should not be called and printed for ZynqMP, need remove
> it.
> 
> Fixes: b500b4a029d57 ("EDAC, synopsys: Add ECC support for ZynqMP DDR
> controller")
> Signed-off-by: Sherry Sun <sherry.sun@nxp.com>

Reviewed-by: Manish Narani <manish.narani@xilinx.com>

Thanks,
Manish
Borislav Petkov March 15, 2020, 7:50 p.m. UTC | #2
On Fri, Feb 28, 2020 at 10:16:23AM +0800, Sherry Sun wrote:
> Since ZynqMP platform call zynqmp_get_error_info() function to get ce/ue
> information. In this function, pinf->col parameter is not used, this
> parameter is only used by Zynq platforme in zynq_get_error_info(). So
> here pinf->col should not be called and printed for ZynqMP, need remove
> it.

Err, the interrupt handler intr_handler() calls the ->get_error_info()
function pointer and then calls handle_error(). I.e, the same path
ending in handle_error() is called on both: Zynq and ZynqMP, one through
the interrupt and the other through the poller.

Because it looks like the interrupt support is only on ZynqMP? I'm
looking at that DDR_ECC_INTR_SUPPORT thing.

If so, then you need to rename the interrupt handler to
zynqmp_intr_handler().

And normal Zynq platform uses check_errors() which then calls
handle_error().

So still not good enough: you probably "fixed" it on ZynqMP but broke
it on Zynq because check_errors() calls ->get_error_info() which is
zynq_get_error_info() and will have populated pinf->col.

You need to differentiate between Zynq and ZynqMP in handle_error() and
issue the proper string depending on the platform.

AFAICT.
Sherry Sun March 16, 2020, 1:30 a.m. UTC | #3
Hi Borislav,

> -----Original Message-----
> From: Borislav Petkov <bp@alien8.de>
> Sent: 2020年3月16日 3:50
> To: Sherry Sun <sherry.sun@nxp.com>
> Cc: mchehab@kernel.org; tony.luck@intel.com; james.morse@arm.com;
> rrichter@marvell.com; michal.simek@xilinx.com; manish.narani@xilinx.com;
> linux-edac@vger.kernel.org; linux-kernel@vger.kernel.org; dl-linux-imx
> <linux-imx@nxp.com>; Frank Li <frank.li@nxp.com>
> Subject: Re: [PATCH v2] EDAC: synopsys: Fix the wrong call of pinf->col
> parameter
> 
> On Fri, Feb 28, 2020 at 10:16:23AM +0800, Sherry Sun wrote:
> > Since ZynqMP platform call zynqmp_get_error_info() function to get
> > ce/ue information. In this function, pinf->col parameter is not used,
> > this parameter is only used by Zynq platforme in
> > zynq_get_error_info(). So here pinf->col should not be called and
> > printed for ZynqMP, need remove it.
> 
> Err, the interrupt handler intr_handler() calls the ->get_error_info() function
> pointer and then calls handle_error(). I.e, the same path ending in
> handle_error() is called on both: Zynq and ZynqMP, one through the
> interrupt and the other through the poller.
> 
> Because it looks like the interrupt support is only on ZynqMP? I'm looking at
> that DDR_ECC_INTR_SUPPORT thing.
> 
> If so, then you need to rename the interrupt handler to
> zynqmp_intr_handler().
> 
> And normal Zynq platform uses check_errors() which then calls
> handle_error().
> 
> So still not good enough: you probably "fixed" it on ZynqMP but broke it on
> Zynq because check_errors() calls ->get_error_info() which is
> zynq_get_error_info() and will have populated pinf->col.
> 
> You need to differentiate between Zynq and ZynqMP in handle_error() and
> issue the proper string depending on the platform.

Maybe you misunderstood the handle_error(), this function has already 
distinguished between Zynq and ZynqMP through priv->p_data->quirks.
And what I am doing in this patch is to remove the useless pinf->col in ZynqMP
part, which won't break the Zynq part.

Thanks
Sherry Sun


> 
> AFAICT.
> 
> --
> Regards/Gruss,
>     Boris.
> 
> https://eur01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fpeopl
> e.kernel.org%2Ftglx%2Fnotes-about-
> netiquette&amp;data=02%7C01%7Csherry.sun%40nxp.com%7C948bf1123f6f
> 4717a61508d7c91a193b%7C686ea1d3bc2b4c6fa92cd99c5c301635%7C0%7C0
> %7C637198986257175149&amp;sdata=O8txcSkTpO3I34KUL0qboQ%2BvPme
> D7lYPHsAhWavevWg%3D&amp;reserved=0
Borislav Petkov March 16, 2020, 10:43 a.m. UTC | #4
On Mon, Mar 16, 2020 at 01:30:39AM +0000, Sherry Sun wrote:
> Maybe you misunderstood the handle_error(), this function has already 
> distinguished between Zynq and ZynqMP through priv->p_data->quirks.
> And what I am doing in this patch is to remove the useless pinf->col in ZynqMP
> part, which won't break the Zynq part.

Ok, I see it now, thanks for clarifying.

Now, in your v2 pls fix this check in handle_error():

	if (!priv->p_data->quirks) {

to *actually* *explicitly* check that the platform for which
this function is called, really supports interrupts, i.e.,
DDR_ECC_INTR_SUPPORT, and not that the ->quirks thing is != 0. Like the
rest of the code does.

Thx.
Sherry Sun March 16, 2020, 12:28 p.m. UTC | #5
Hi Borislav,

> -----Original Message-----
> From: Borislav Petkov <bp@alien8.de>
> Sent: 2020年3月16日 18:44
> To: Sherry Sun <sherry.sun@nxp.com>
> Cc: mchehab@kernel.org; tony.luck@intel.com; james.morse@arm.com;
> rrichter@marvell.com; michal.simek@xilinx.com; manish.narani@xilinx.com;
> linux-edac@vger.kernel.org; linux-kernel@vger.kernel.org; dl-linux-imx
> <linux-imx@nxp.com>; Frank Li <frank.li@nxp.com>
> Subject: Re: [PATCH v2] EDAC: synopsys: Fix the wrong call of pinf->col
> parameter
> 
> On Mon, Mar 16, 2020 at 01:30:39AM +0000, Sherry Sun wrote:
> > Maybe you misunderstood the handle_error(), this function has already
> > distinguished between Zynq and ZynqMP through priv->p_data->quirks.
> > And what I am doing in this patch is to remove the useless pinf->col
> > in ZynqMP part, which won't break the Zynq part.
> 
> Ok, I see it now, thanks for clarifying.
> 
> Now, in your v2 pls fix this check in handle_error():
> 
> 	if (!priv->p_data->quirks) {
> 
> to *actually* *explicitly* check that the platform for which this function is
> called, really supports interrupts, i.e., DDR_ECC_INTR_SUPPORT, and not that
> the ->quirks thing is != 0. Like the rest of the code does.
> 

Ok, I will fix this check.

Best regards
Sherry Sun

> Thx.
> 
> --
> Regards/Gruss,
>     Boris.
> 
> https://eur01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fpeopl
> e.kernel.org%2Ftglx%2Fnotes-about-
> netiquette&amp;data=02%7C01%7Csherry.sun%40nxp.com%7Ccd843239ffd
> d48a05ce408d7c996ebd5%7C686ea1d3bc2b4c6fa92cd99c5c301635%7C0%7C
> 0%7C637199522356909838&amp;sdata=5p6JuXQQJgyjlaMOUH1cLLSDFbUzM
> R%2Fjzs5HC%2FeU1Zc%3D&amp;reserved=0
diff mbox series

Patch

diff --git a/drivers/edac/synopsys_edac.c b/drivers/edac/synopsys_edac.c
index 880ffd833718..da707d539f7f 100644
--- a/drivers/edac/synopsys_edac.c
+++ b/drivers/edac/synopsys_edac.c
@@ -484,8 +484,8 @@  static void handle_error(struct mem_ctl_info *mci, struct synps_ecc_status *p)
 				 pinf->bitpos, pinf->data);
 		} else {
 			snprintf(priv->message, SYNPS_EDAC_MSG_SIZE,
-				 "DDR ECC error type:%s Row %d Bank %d Col %d BankGroup Number %d Block Number %d Bit Position: %d Data: 0x%08x",
-				 "CE", pinf->row, pinf->bank, pinf->col,
+				 "DDR ECC error type:%s Row %d Bank %d BankGroup Number %d Block Number %d Bit Position: %d Data: 0x%08x",
+				 "CE", pinf->row, pinf->bank,
 				 pinf->bankgrpnr, pinf->blknr,
 				 pinf->bitpos, pinf->data);
 		}
@@ -503,8 +503,8 @@  static void handle_error(struct mem_ctl_info *mci, struct synps_ecc_status *p)
 				"UE", pinf->row, pinf->bank, pinf->col);
 		} else {
 			snprintf(priv->message, SYNPS_EDAC_MSG_SIZE,
-				 "DDR ECC error type :%s Row %d Bank %d Col %d BankGroup Number %d Block Number %d",
-				 "UE", pinf->row, pinf->bank, pinf->col,
+				 "DDR ECC error type :%s Row %d Bank %d BankGroup Number %d Block Number %d",
+				 "UE", pinf->row, pinf->bank,
 				 pinf->bankgrpnr, pinf->blknr);
 		}