diff for duplicates of <200802181100.12995.ossthema@de.ibm.com> diff --git a/a/1.txt b/N1/1.txt index ab5d7e9..f4c7a3f 100644 --- a/a/1.txt +++ b/N1/1.txt @@ -3,38 +3,36 @@ switching to proper mail client... Dave Hansen <haveblue@us.ibm.com> wrote on 15.02.2008 17:55:38: > I've been thinking about that, and I don't think you really *need* to -> keep a comprehensive map like that.=20 ->=20 +> keep a comprehensive map like that. +> > When the memory is in a particular configuration (range of memory > present along with unique set of holes) you get a unique ehea_bmap -> configuration. =A0That layout is completely predictable. ->=20 +> configuration. That layout is completely predictable. +> > So, if at any time you want to figure out what the ehea_bmap address for > a particular *Linux* virtual address is, you just need to pretend that > you're creating the entire ehea_bmap, use the same algorithm and figure > out host you would have placed things, and use that result. ->=20 +> > Now, that's going to be a slow, crappy linear search (but maybe not as -> slow as recreating the silly thing). =A0So, you might eventually run into -> some scalability problems with a lot of packets going around. =A0But, I'd +> slow as recreating the silly thing). So, you might eventually run into +> some scalability problems with a lot of packets going around. But, I'd > be curious if you do in practice. -Up to 14 addresses translation per packet (sg_list) might be required on=20 -the transmit side. On receive side it is only 1. Most packets require only= -=20 -very few translations (1 or sometimes more) =A0translations. However,=20 -with more then 700.000 packets per second this approach does not seem=20 +Up to 14 addresses translation per packet (sg_list) might be required on +the transmit side. On receive side it is only 1. Most packets require only +very few translations (1 or sometimes more) translations. However, +with more then 700.000 packets per second this approach does not seem reasonable from performance perspective when memory is fragmented as you described. ->=20 +> > The other idea is that you create a mapping that is precisely 1:1 with -> kernel memory. =A0Let's say you have two sections present, 0 and 100. =A0= -You +> kernel memory. Let's say you have two sections present, 0 and 100. You > have a high_section_index of 100, and you vmalloc() a 100 entry array. ->=20 -> You need to create a *CONTIGUOUS* ehea map? =A0Create one like this: ->=20 +> +> You need to create a *CONTIGUOUS* ehea map? Create one like this: +> > EHEA_VADDR->Linux Section > 0->0 > 1->0 @@ -42,18 +40,17 @@ You > 3->0 > ... > 100->100 ->=20 -> It's contiguous. =A0Each area points to a valid Linux memory address. +> +> It's contiguous. Each area points to a valid Linux memory address. > It's also discernable in O(1) to what EHEA address a given Linux address -> is mapped. =A0You just have a couple of duplicate entries.=20 +> is mapped. You just have a couple of duplicate entries. -This has a serious issues with constraint I mentions in the previous mail:= -=20 +This has a serious issues with constraint I mentions in the previous mail: "- MRs can have a maximum size of the memory available under linux" -The requirement is not met that the memory region must not be=20 -larger then the available memory for that partition. The "create MR"=20 +The requirement is not met that the memory region must not be +larger then the available memory for that partition. The "create MR" H_CALL will fails (we tried this and discussed with FW development) diff --git a/a/content_digest b/N1/content_digest index cd45eb9..d0d574d 100644 --- a/a/content_digest +++ b/N1/content_digest @@ -5,14 +5,14 @@ "Subject\0Re: [PATCH] drivers/base: export gpl (un)register_memory_notifier\0" "Date\0Mon, 18 Feb 2008 11:00:11 +0100\0" "To\0linuxppc-dev@ozlabs.org\0" - "Cc\0Thomas Q Klein <TKLEIN@de.ibm.com>" + "Cc\0Dave Hansen <haveblue@us.ibm.com>" + Christoph Raisch <RAISCH@de.ibm.com> + Thomas Q Klein <TKLEIN@de.ibm.com> ossthema@linux.vnet.ibm.com Jan-Bernd Themann <THEMANN@de.ibm.com> Greg KH <greg@kroah.com> - Dave Hansen <haveblue@us.ibm.com> apw <apw@uk.ibm.com> linux-kernel <linux-kernel@vger.kernel.org> - Christoph Raisch <RAISCH@de.ibm.com> Badari Pulavarty <pbadari@us.ibm.com> netdev <netdev@vger.kernel.org> " tklein@linux.ibm.com\0" @@ -23,38 +23,36 @@ "Dave Hansen <haveblue@us.ibm.com> wrote on 15.02.2008 17:55:38:\n" "\n" "> I've been thinking about that, and I don't think you really *need* to\n" - "> keep a comprehensive map like that.=20\n" - ">=20\n" + "> keep a comprehensive map like that. \n" + "> \n" "> When the memory is in a particular configuration (range of memory\n" "> present along with unique set of holes) you get a unique ehea_bmap\n" - "> configuration. =A0That layout is completely predictable.\n" - ">=20\n" + "> configuration. \302\240That layout is completely predictable.\n" + "> \n" "> So, if at any time you want to figure out what the ehea_bmap address for\n" "> a particular *Linux* virtual address is, you just need to pretend that\n" "> you're creating the entire ehea_bmap, use the same algorithm and figure\n" "> out host you would have placed things, and use that result.\n" - ">=20\n" + "> \n" "> Now, that's going to be a slow, crappy linear search (but maybe not as\n" - "> slow as recreating the silly thing). =A0So, you might eventually run into\n" - "> some scalability problems with a lot of packets going around. =A0But, I'd\n" + "> slow as recreating the silly thing). \302\240So, you might eventually run into\n" + "> some scalability problems with a lot of packets going around. \302\240But, I'd\n" "> be curious if you do in practice.\n" "\n" - "Up to 14 addresses translation per packet (sg_list) might be required on=20\n" - "the transmit side. On receive side it is only 1. Most packets require only=\n" - "=20\n" - "very few translations (1 or sometimes more) =A0translations. However,=20\n" - "with more then 700.000 packets per second this approach does not seem=20\n" + "Up to 14 addresses translation per packet (sg_list) might be required on \n" + "the transmit side. On receive side it is only 1. Most packets require only \n" + "very few translations (1 or sometimes more) \302\240translations. However, \n" + "with more then 700.000 packets per second this approach does not seem \n" "reasonable from performance perspective when memory is fragmented as you\n" "described.\n" "\n" - ">=20\n" + "> \n" "> The other idea is that you create a mapping that is precisely 1:1 with\n" - "> kernel memory. =A0Let's say you have two sections present, 0 and 100. =A0=\n" - "You\n" + "> kernel memory. \302\240Let's say you have two sections present, 0 and 100. \302\240You\n" "> have a high_section_index of 100, and you vmalloc() a 100 entry array.\n" - ">=20\n" - "> You need to create a *CONTIGUOUS* ehea map? =A0Create one like this:\n" - ">=20\n" + "> \n" + "> You need to create a *CONTIGUOUS* ehea map? \302\240Create one like this:\n" + "> \n" "> EHEA_VADDR->Linux Section\n" "> 0->0\n" "> 1->0\n" @@ -62,22 +60,21 @@ "> 3->0\n" "> ...\n" "> 100->100\n" - ">=20\n" - "> It's contiguous. =A0Each area points to a valid Linux memory address.\n" + "> \n" + "> It's contiguous. \302\240Each area points to a valid Linux memory address.\n" "> It's also discernable in O(1) to what EHEA address a given Linux address\n" - "> is mapped. =A0You just have a couple of duplicate entries.=20\n" + "> is mapped. \302\240You just have a couple of duplicate entries. \n" "\n" - "This has a serious issues with constraint I mentions in the previous mail:=\n" - "=20\n" + "This has a serious issues with constraint I mentions in the previous mail: \n" "\n" "\"- MRs can have a maximum size of the memory available under linux\"\n" "\n" - "The requirement is not met that the memory region must not be=20\n" - "larger then the available memory for that partition. The \"create MR\"=20\n" + "The requirement is not met that the memory region must not be \n" + "larger then the available memory for that partition. The \"create MR\" \n" "H_CALL will fails (we tried this and discussed with FW development)\n" "\n" "\n" "Regards,\n" Jan-Bernd & Christoph -fc0b27d8ab89a019218684371202d9cf52dc63de2ee5209dd1d737ea831fc149 +34833f875b235a08142e65982da66cf298588041f71446c09eeb85307579fcb2
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.