diff for duplicates of <1493932034.25397.37.camel@buserror.net> diff --git a/a/1.txt b/N1/1.txt index cc8f689..90d21ed 100644 --- a/a/1.txt +++ b/N1/1.txt @@ -1,19 +1,19 @@ On Thu, 2017-05-04 at 06:58 +0200, Karim Eshapa wrote: > + stop = jiffies + 10000; > + /* -> + * if MR was full and h/w had other FQRNI entries to produce, we -> + * need to allow it time to produce those entries once the -> + * existing entries are consumed. A worst-case situation -> + * (fully-loaded system) means h/w sequencers may have to do 3-4 -> + * other things before servicing the portal's MR pump, each of -> + * which (if slow) may take ~50 qman cycles (which is ~200 -> + * processor cycles). So rounding up and then multiplying this -> + * worst-case estimate by a factor of 10, just to be -> + * ultra-paranoid, goes as high as 10,000 cycles. NB, we consume -> + * one entry at a time, so h/w has an opportunity to produce new -> + * entries well before the ring has been fully consumed, so -> + * we're being *really* paranoid here. -> + */ +> + ?* if MR was full and h/w had other FQRNI entries to produce, we +> + ?* need to allow it time to produce those entries once the +> + ?* existing entries are consumed. A worst-case situation +> + ?* (fully-loaded system) means h/w sequencers may have to do 3-4 +> + ?* other things before servicing the portal's MR pump, each of +> + ?* which (if slow) may take ~50 qman cycles (which is ~200 +> + ?* processor cycles). So rounding up and then multiplying this +> + ?* worst-case estimate by a factor of 10, just to be +> + ?* ultra-paranoid, goes as high as 10,000 cycles. NB, we consume +> + ?* one entry at a time, so h/w has an opportunity to produce new +> + ?* entries well before the ring has been fully consumed, so +> + ?* we're being *really* paranoid here. +> + ?*/ OK, upon reading this more closely it seems the intent was to delay for 10,000 *processor cycles* and somehow that got turned into 10,000 jiffies (which is diff --git a/a/content_digest b/N1/content_digest index 57b2461..22d3c9f 100644 --- a/a/content_digest +++ b/N1/content_digest @@ -1,33 +1,27 @@ "ref\01493508775.25397.26.camel@buserror.net\0" "ref\01493873917-7484-1-git-send-email-karim.eshapa@gmail.com\0" - "From\0Scott Wood <oss@buserror.net>\0" - "Subject\0Re: [PATCH v2] drivers:soc:fsl:qbman:qman.c: Sleep instead of stuck hacking jiffies.\0" + "From\0oss@buserror.net (Scott Wood)\0" + "Subject\0[PATCH v2] drivers:soc:fsl:qbman:qman.c: Sleep instead of stuck hacking jiffies.\0" "Date\0Thu, 04 May 2017 16:07:14 -0500\0" - "To\0Karim Eshapa <karim.eshapa@gmail.com>\0" - "Cc\0claudiu.manoil@nxp.com" - roy.pledge@nxp.com - colin.king@canonical.com - linuxppc-dev@lists.ozlabs.org - linux-arm-kernel@lists.infradead.org - " linux-kernel@vger.kernel.org\0" + "To\0linux-arm-kernel@lists.infradead.org\0" "\00:1\0" "b\0" "On Thu, 2017-05-04 at 06:58 +0200, Karim Eshapa wrote:\n" "> +\tstop = jiffies + 10000;\n" "> +\t/*\n" - "> +\t\302\240* if MR was full and h/w had other FQRNI entries to produce, we\n" - "> +\t\302\240* need to allow it time to produce those entries once the\n" - "> +\t\302\240* existing entries are consumed. A worst-case situation\n" - "> +\t\302\240* (fully-loaded system) means h/w sequencers may have to do 3-4\n" - "> +\t\302\240* other things before servicing the portal's MR pump, each of\n" - "> +\t\302\240* which (if slow) may take ~50 qman cycles (which is ~200\n" - "> +\t\302\240* processor cycles). So rounding up and then multiplying this\n" - "> +\t\302\240* worst-case estimate by a factor of 10, just to be\n" - "> +\t\302\240* ultra-paranoid, goes as high as 10,000 cycles. NB, we consume\n" - "> +\t\302\240* one entry at a time, so h/w has an opportunity to produce new\n" - "> +\t\302\240* entries well before the ring has been fully consumed, so\n" - "> +\t\302\240* we're being *really* paranoid here.\n" - "> +\t\302\240*/\n" + "> +\t?* if MR was full and h/w had other FQRNI entries to produce, we\n" + "> +\t?* need to allow it time to produce those entries once the\n" + "> +\t?* existing entries are consumed. A worst-case situation\n" + "> +\t?* (fully-loaded system) means h/w sequencers may have to do 3-4\n" + "> +\t?* other things before servicing the portal's MR pump, each of\n" + "> +\t?* which (if slow) may take ~50 qman cycles (which is ~200\n" + "> +\t?* processor cycles). So rounding up and then multiplying this\n" + "> +\t?* worst-case estimate by a factor of 10, just to be\n" + "> +\t?* ultra-paranoid, goes as high as 10,000 cycles. NB, we consume\n" + "> +\t?* one entry at a time, so h/w has an opportunity to produce new\n" + "> +\t?* entries well before the ring has been fully consumed, so\n" + "> +\t?* we're being *really* paranoid here.\n" + "> +\t?*/\n" "\n" "OK, upon reading this more closely it seems the intent was to delay for 10,000\n" "*processor cycles* and somehow that got turned into 10,000 jiffies (which is\n" @@ -38,4 +32,4 @@ "\n" -Scott -0e9f2ec4c11bf2b961420adeca888eadae14301517461511c37681047486523e +237681a6975914da3d7a7c33dd35a224ed50f2955b6ab818e5dd79f392f6e8a8
This is an external index of several public inboxes, see mirroring instructions on how to clone and mirror all data and code used by this external index.