From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from smtp1.linuxfoundation.org (smtp1.linux-foundation.org [172.17.192.35]) by mail.linuxfoundation.org (Postfix) with ESMTPS id CC2DB410 for ; Thu, 30 Jul 2015 13:31:43 +0000 (UTC) Received: from casper.infradead.org (casper.infradead.org [85.118.1.10]) by smtp1.linuxfoundation.org (Postfix) with ESMTPS id 2D33815A for ; Thu, 30 Jul 2015 13:31:42 +0000 (UTC) Message-ID: <1438263098.26511.179.camel@infradead.org> From: David Woodhouse To: Joerg Roedel , ksummit-discuss@lists.linuxfoundation.org Date: Thu, 30 Jul 2015 14:31:38 +0100 In-Reply-To: <20150730130027.GA14980@8bytes.org> References: <20150730130027.GA14980@8bytes.org> Content-Type: multipart/signed; micalg="sha-1"; protocol="application/x-pkcs7-signature"; boundary="=-lBQzFsK0B7j1nNmLRgp1" Mime-Version: 1.0 Subject: Re: [Ksummit-discuss] [CORE TOPIC] Core Kernel support for Compute-Offload Devices List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , --=-lBQzFsK0B7j1nNmLRgp1 Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: quoted-printable On Thu, 2015-07-30 at 15:00 +0200, Joerg Roedel wrote: > [ > The topic is highly technical and could be a tech topic. But it also > touches multiple subsystems, so I decided to submit it as a core > topic. > ] >=20 > Across architectures and vendors there are new devices coming up for > offloading tasks from the CPUs. Most of these devices are capable to > operate on user address spaces. >=20 > Besides the commonalities there are important differences in the=20 > memory > model these devices offer. Some work only on system RAM, others come > with their own memory which may or may not be accessible by the CPU. >=20 > I'd like to discuss what support we need in the core kernel for these > devices. A probably incomplete list of open questions: >=20 > (1) Do we need the concept of an off-CPU task in the kernel > together with a common interface to create and manage them > and probably a (collection of) batch scheduler(s) for these > tasks? >=20 > (2) Changes in memory management for devices accessing user > address spaces: > =20 > (2.1) How can we best support the different memory models > these devices support? > =20 > (2.2) How do we handle the off-CPU users of an mm_struct? > =20 > (2.3) How can we attach common state for off-CPU tasks to > mm_struct (and what needs to be in there)? And how do we handle the assignment of Address Space IDs? The AMD implementation currently allows the PASID space to be managed per -device, but I understand ARM systems handle the TLB shootdown broadcasts in hardware and need the PASID that the device sees to be identical to the ASID on the CPU's MMU? And there are reasons why we might actually want that model on Intel systems too. I'm working on the Intel SVM right now, and looking at a single-PASID-space model (partly because the PASID tables have to be physically contiguous, and they can be huge!). > (3) Does it make sense to implement automatic migration of > system memory to device memory (when available) and vice > versa? How do we decide what and when to migrate? This is quite a horrid one, but perhaps ties into generic NUMA considerations =E2=80=94 if a memory page is being frequently accessed by something that it's far away from, can we move it to closer memory?=20 The question is how we handle that. We do have Extended Accessed bits in the Intel implementation of SVM that let us know that a given PTE was used from a device. Although not *which* device, in cases where there might be more than one. > (4) What features do we require in the hardware to support it > with a common interface? >=20 > I think it would be great if the kernel would have a common interface > for these kind of devices. Currently every vendor develops its own > interface with various hacks to work around core code behavior. Right. For now it's almost all internal on-chip stuff so it's kind of tolerable to have vendor-specific implenentations. But we are starting to see PCIe root ports which support the necessary TLP prefixes to support SVM on discrete devices. And then it'll be really important to have this working cross-platform. --=20 dwmw2 --=-lBQzFsK0B7j1nNmLRgp1 Content-Type: application/x-pkcs7-signature; name="smime.p7s" Content-Disposition: attachment; filename="smime.p7s" Content-Transfer-Encoding: base64 MIAGCSqGSIb3DQEHAqCAMIACAQExCzAJBgUrDgMCGgUAMIAGCSqGSIb3DQEHAQAAoIISjjCCBicw ggUPoAMCAQICAw3vNzANBgkqhkiG9w0BAQUFADCBjDELMAkGA1UEBhMCSUwxFjAUBgNVBAoTDVN0 YXJ0Q29tIEx0ZC4xKzApBgNVBAsTIlNlY3VyZSBEaWdpdGFsIENlcnRpZmljYXRlIFNpZ25pbmcx ODA2BgNVBAMTL1N0YXJ0Q29tIENsYXNzIDEgUHJpbWFyeSBJbnRlcm1lZGlhdGUgQ2xpZW50IENB MB4XDTE1MDUwNTA5NDM0MVoXDTE2MDUwNTA5NTMzNlowQjEcMBoGA1UEAwwTZHdtdzJAaW5mcmFk ZWFkLm9yZzEiMCAGCSqGSIb3DQEJARYTZHdtdzJAaW5mcmFkZWFkLm9yZzCCASIwDQYJKoZIhvcN AQEBBQADggEPADCCAQoCggEBAMkbm9kPbx1j/X4RVyf/pPKSYwelcco69TvnQQbKM8m8xkWjXJI1 jpJ1jMaGUZGFToINMSZi7lZawUozudWbXSKy1SikENSTJHffsdRAIlsp+hR8vWvjsKUry6sEdqPG doa5RY7+N4WRusWZDYW/RRWE6i9EL9qV86CVPYqw22UBOUw4/j/HVGCV6TSB8yE5iEwhk/hUuzRr FZm1MJMR7mCS7BCR8Lr5jFY61lWpBiXNXIxLZCvDc26KR5L5tYX43iUVO3fzES1GRVoYnxxk2tmz fcsZG5vK+Trc9L8OZJfkYrEHH3+Iw41MQ0w/djVtYr1+HYldx0QmYXAtnhIj+UMCAwEAAaOCAtkw ggLVMAkGA1UdEwQCMAAwCwYDVR0PBAQDAgSwMB0GA1UdJQQWMBQGCCsGAQUFBwMCBggrBgEFBQcD BDAdBgNVHQ4EFgQUszC96C3w5/2+d+atSr0IpT26YI4wHwYDVR0jBBgwFoAUU3Ltkpzg2ssBXHx+ ljVO8tS4UYIwHgYDVR0RBBcwFYETZHdtdzJAaW5mcmFkZWFkLm9yZzCCAUwGA1UdIASCAUMwggE/ MIIBOwYLKwYBBAGBtTcBAgMwggEqMC4GCCsGAQUFBwIBFiJodHRwOi8vd3d3LnN0YXJ0c3NsLmNv bS9wb2xpY3kucGRmMIH3BggrBgEFBQcCAjCB6jAnFiBTdGFydENvbSBDZXJ0aWZpY2F0aW9uIEF1 dGhvcml0eTADAgEBGoG+VGhpcyBjZXJ0aWZpY2F0ZSB3YXMgaXNzdWVkIGFjY29yZGluZyB0byB0 aGUgQ2xhc3MgMSBWYWxpZGF0aW9uIHJlcXVpcmVtZW50cyBvZiB0aGUgU3RhcnRDb20gQ0EgcG9s aWN5LCByZWxpYW5jZSBvbmx5IGZvciB0aGUgaW50ZW5kZWQgcHVycG9zZSBpbiBjb21wbGlhbmNl IG9mIHRoZSByZWx5aW5nIHBhcnR5IG9ibGlnYXRpb25zLjA2BgNVHR8ELzAtMCugKaAnhiVodHRw Oi8vY3JsLnN0YXJ0c3NsLmNvbS9jcnR1MS1jcmwuY3JsMIGOBggrBgEFBQcBAQSBgTB/MDkGCCsG AQUFBzABhi1odHRwOi8vb2NzcC5zdGFydHNzbC5jb20vc3ViL2NsYXNzMS9jbGllbnQvY2EwQgYI KwYBBQUHMAKGNmh0dHA6Ly9haWEuc3RhcnRzc2wuY29tL2NlcnRzL3N1Yi5jbGFzczEuY2xpZW50 LmNhLmNydDAjBgNVHRIEHDAahhhodHRwOi8vd3d3LnN0YXJ0c3NsLmNvbS8wDQYJKoZIhvcNAQEF BQADggEBAHMQmxHHodpS85X8HRyxhvfkys7r+taCNOaNU9cxQu/cZ/6k5nS2qGNMzZ6jb7ueY/V7 7p+4DW/9ZWODDTf4Fz00mh5SSVc20Bz7t+hhxwHd62PZgENh5i76Qq2tw48U8AsYo5damHby1epf neZafLpUkLLO7AGBJIiRVTevdvyXQ0qnixOmKMWyvrhSNGuVIKVdeqLP+102Dwf+dpFyw+j1hz28 jEEKpHa+NR1b2kXuSPi/rMGhexwlJOh4tK8KQ6Ryr0rIN//NSbOgbyYZrzc/ZUWX9V5OA84ChFb2 vkFl0OcYrttp/rhDBLITwffPxSZeoBh9H7zYzkbCXKL3BUIwggYnMIIFD6ADAgECAgMN7zcwDQYJ KoZIhvcNAQEFBQAwgYwxCzAJBgNVBAYTAklMMRYwFAYDVQQKEw1TdGFydENvbSBMdGQuMSswKQYD VQQLEyJTZWN1cmUgRGlnaXRhbCBDZXJ0aWZpY2F0ZSBTaWduaW5nMTgwNgYDVQQDEy9TdGFydENv bSBDbGFzcyAxIFByaW1hcnkgSW50ZXJtZWRpYXRlIENsaWVudCBDQTAeFw0xNTA1MDUwOTQzNDFa Fw0xNjA1MDUwOTUzMzZaMEIxHDAaBgNVBAMME2R3bXcyQGluZnJhZGVhZC5vcmcxIjAgBgkqhkiG 9w0BCQEWE2R3bXcyQGluZnJhZGVhZC5vcmcwggEiMA0GCSqGSIb3DQEBAQUAA4IBDwAwggEKAoIB AQDJG5vZD28dY/1+EVcn/6TykmMHpXHKOvU750EGyjPJvMZFo1ySNY6SdYzGhlGRhU6CDTEmYu5W WsFKM7nVm10istUopBDUkyR337HUQCJbKfoUfL1r47ClK8urBHajxnaGuUWO/jeFkbrFmQ2Fv0UV hOovRC/alfOglT2KsNtlATlMOP4/x1Rglek0gfMhOYhMIZP4VLs0axWZtTCTEe5gkuwQkfC6+YxW OtZVqQYlzVyMS2Qrw3NuikeS+bWF+N4lFTt38xEtRkVaGJ8cZNrZs33LGRubyvk63PS/DmSX5GKx Bx9/iMONTENMP3Y1bWK9fh2JXcdEJmFwLZ4SI/lDAgMBAAGjggLZMIIC1TAJBgNVHRMEAjAAMAsG A1UdDwQEAwIEsDAdBgNVHSUEFjAUBggrBgEFBQcDAgYIKwYBBQUHAwQwHQYDVR0OBBYEFLMwvegt 8Of9vnfmrUq9CKU9umCOMB8GA1UdIwQYMBaAFFNy7ZKc4NrLAVx8fpY1TvLUuFGCMB4GA1UdEQQX MBWBE2R3bXcyQGluZnJhZGVhZC5vcmcwggFMBgNVHSAEggFDMIIBPzCCATsGCysGAQQBgbU3AQID MIIBKjAuBggrBgEFBQcCARYiaHR0cDovL3d3dy5zdGFydHNzbC5jb20vcG9saWN5LnBkZjCB9wYI KwYBBQUHAgIwgeowJxYgU3RhcnRDb20gQ2VydGlmaWNhdGlvbiBBdXRob3JpdHkwAwIBARqBvlRo aXMgY2VydGlmaWNhdGUgd2FzIGlzc3VlZCBhY2NvcmRpbmcgdG8gdGhlIENsYXNzIDEgVmFsaWRh dGlvbiByZXF1aXJlbWVudHMgb2YgdGhlIFN0YXJ0Q29tIENBIHBvbGljeSwgcmVsaWFuY2Ugb25s eSBmb3IgdGhlIGludGVuZGVkIHB1cnBvc2UgaW4gY29tcGxpYW5jZSBvZiB0aGUgcmVseWluZyBw YXJ0eSBvYmxpZ2F0aW9ucy4wNgYDVR0fBC8wLTAroCmgJ4YlaHR0cDovL2NybC5zdGFydHNzbC5j b20vY3J0dTEtY3JsLmNybDCBjgYIKwYBBQUHAQEEgYEwfzA5BggrBgEFBQcwAYYtaHR0cDovL29j c3Auc3RhcnRzc2wuY29tL3N1Yi9jbGFzczEvY2xpZW50L2NhMEIGCCsGAQUFBzAChjZodHRwOi8v YWlhLnN0YXJ0c3NsLmNvbS9jZXJ0cy9zdWIuY2xhc3MxLmNsaWVudC5jYS5jcnQwIwYDVR0SBBww GoYYaHR0cDovL3d3dy5zdGFydHNzbC5jb20vMA0GCSqGSIb3DQEBBQUAA4IBAQBzEJsRx6HaUvOV /B0csYb35MrO6/rWgjTmjVPXMULv3Gf+pOZ0tqhjTM2eo2+7nmP1e+6fuA1v/WVjgw03+Bc9NJoe UklXNtAc+7foYccB3etj2YBDYeYu+kKtrcOPFPALGKOXWph28tXqX53mWny6VJCyzuwBgSSIkVU3 r3b8l0NKp4sTpijFsr64UjRrlSClXXqiz/tdNg8H/naRcsPo9Yc9vIxBCqR2vjUdW9pF7kj4v6zB oXscJSToeLSvCkOkcq9KyDf/zUmzoG8mGa83P2VFl/VeTgPOAoRW9r5BZdDnGK7baf64QwSyE8H3 z8UmXqAYfR+82M5Gwlyi9wVCMIIGNDCCBBygAwIBAgIBHjANBgkqhkiG9w0BAQUFADB9MQswCQYD VQQGEwJJTDEWMBQGA1UEChMNU3RhcnRDb20gTHRkLjErMCkGA1UECxMiU2VjdXJlIERpZ2l0YWwg Q2VydGlmaWNhdGUgU2lnbmluZzEpMCcGA1UEAxMgU3RhcnRDb20gQ2VydGlmaWNhdGlvbiBBdXRo b3JpdHkwHhcNMDcxMDI0MjEwMTU1WhcNMTcxMDI0MjEwMTU1WjCBjDELMAkGA1UEBhMCSUwxFjAU BgNVBAoTDVN0YXJ0Q29tIEx0ZC4xKzApBgNVBAsTIlNlY3VyZSBEaWdpdGFsIENlcnRpZmljYXRl IFNpZ25pbmcxODA2BgNVBAMTL1N0YXJ0Q29tIENsYXNzIDEgUHJpbWFyeSBJbnRlcm1lZGlhdGUg Q2xpZW50IENBMIIBIjANBgkqhkiG9w0BAQEFAAOCAQ8AMIIBCgKCAQEAxwmDzM4t2BqxKaQuE6uW vooyg4ymiEGWVUet1G8SD+rqvyNH4QrvnEIaFHxOhESip7vMz39ScLpNLbL1QpOlPW/tFIzNHS3q d2XRNYG5Sv9RcGE+T4qbLtsjjJbi6sL7Ls/f/X9ftTyhxvxWkf8KW37iKrueKsxw2HqolH7GM6FX 5UfNAwAu4ZifkpmZzU1slBhyWwaQPEPPZRsWoTb7q8hmgv6Nv3Hg9rmA1/VPBIOQ6SKRkHXG0Hhm q1dOFoAFI411+a/9nWm5rcVjGcIWZ2v/43Yksq60jExipA4l5uv9/+Hm33mbgmCszdj/Dthf13tg Av2O83hLJ0exTqfrlwIDAQABo4IBrTCCAakwDwYDVR0TAQH/BAUwAwEB/zAOBgNVHQ8BAf8EBAMC AQYwHQYDVR0OBBYEFFNy7ZKc4NrLAVx8fpY1TvLUuFGCMB8GA1UdIwQYMBaAFE4L7xqkQFulF2mH MMo0aEPQQa7yMGYGCCsGAQUFBwEBBFowWDAnBggrBgEFBQcwAYYbaHR0cDovL29jc3Auc3RhcnRz c2wuY29tL2NhMC0GCCsGAQUFBzAChiFodHRwOi8vd3d3LnN0YXJ0c3NsLmNvbS9zZnNjYS5jcnQw WwYDVR0fBFQwUjAnoCWgI4YhaHR0cDovL3d3dy5zdGFydHNzbC5jb20vc2ZzY2EuY3JsMCegJaAj hiFodHRwOi8vY3JsLnN0YXJ0c3NsLmNvbS9zZnNjYS5jcmwwgYAGA1UdIAR5MHcwdQYLKwYBBAGB tTcBAgEwZjAuBggrBgEFBQcCARYiaHR0cDovL3d3dy5zdGFydHNzbC5jb20vcG9saWN5LnBkZjA0 BggrBgEFBQcCARYoaHR0cDovL3d3dy5zdGFydHNzbC5jb20vaW50ZXJtZWRpYXRlLnBkZjANBgkq hkiG9w0BAQUFAAOCAgEACoMIfXirLAZcuGOMXq4cuSN3TaFx2H2GvD5VSy/6rV55BYHbWNaPeQn3 oBSU8KgQZn/Kck1JxbLpAxVCNtsxeW1R87ifhsYZ0qjdrA9anrW2MAWCtosmAOT4OxK9QPoSjCMx M3HbkZCDJgnlE8jMopH21BbyAYr7b5EfGRQJNtgWcvqSXwKHnTutR08+Kkn0KAkXCzeQNLeA5LlY UzFyM7kPAp8pIRMQ+seHunmyG642S2+y/qHEdMuGIwpfz3eDF1PdctL04qYK/zu+Qg1Bw0RwgigV Zs/0c5HP2/e9DBHh7eSwtzYlk4AUr6yxLlcwSjOfOmKEQ/Q8tzh0IFiNu9IPuTGAPBn4CPxD0+Ru 8T2wg8/s43R/PT3kd1OEqOJUl7q+h+r6fpvU0Fzxd2tC8Ga6fDEPme+1Nbi+03pVjuZQKbGwKJ66 gEn06WqaxVZC+J8hh/jR0k9mST1iAZPNYulcNJ8tKmVtjYsv0L1TSm2+NwON58tO+pIVzu3DWwSE XSf+qkDavQam+QtEOZxLBXI++aMUEapSn+k3Lxm48ZCYfAWLb/Xj7F5JQMbZvCexglAbYR0kIHqW 5DnsYSdMD/IplJMojx0NBrxJ3fN9dvX2Y6BIXRsF1du4qESm4/3CKuyUV7p9DW3mPlHTGLvYxnyK Qy7VFBkoLINszBrOUeIxggNvMIIDawIBATCBlDCBjDELMAkGA1UEBhMCSUwxFjAUBgNVBAoTDVN0 YXJ0Q29tIEx0ZC4xKzApBgNVBAsTIlNlY3VyZSBEaWdpdGFsIENlcnRpZmljYXRlIFNpZ25pbmcx ODA2BgNVBAMTL1N0YXJ0Q29tIENsYXNzIDEgUHJpbWFyeSBJbnRlcm1lZGlhdGUgQ2xpZW50IENB AgMN7zcwCQYFKw4DAhoFAKCCAa8wGAYJKoZIhvcNAQkDMQsGCSqGSIb3DQEHATAcBgkqhkiG9w0B CQUxDxcNMTUwNzMwMTMzMTM4WjAjBgkqhkiG9w0BCQQxFgQUVpmet7w7bY367cCV0tNqq3leLEUw gaUGCSsGAQQBgjcQBDGBlzCBlDCBjDELMAkGA1UEBhMCSUwxFjAUBgNVBAoTDVN0YXJ0Q29tIEx0 ZC4xKzApBgNVBAsTIlNlY3VyZSBEaWdpdGFsIENlcnRpZmljYXRlIFNpZ25pbmcxODA2BgNVBAMT L1N0YXJ0Q29tIENsYXNzIDEgUHJpbWFyeSBJbnRlcm1lZGlhdGUgQ2xpZW50IENBAgMN7zcwgacG CyqGSIb3DQEJEAILMYGXoIGUMIGMMQswCQYDVQQGEwJJTDEWMBQGA1UEChMNU3RhcnRDb20gTHRk LjErMCkGA1UECxMiU2VjdXJlIERpZ2l0YWwgQ2VydGlmaWNhdGUgU2lnbmluZzE4MDYGA1UEAxMv U3RhcnRDb20gQ2xhc3MgMSBQcmltYXJ5IEludGVybWVkaWF0ZSBDbGllbnQgQ0ECAw3vNzANBgkq hkiG9w0BAQEFAASCAQDD3Ue/oageUPKG3onh2qHkST1SckCuiK1FBsUalGjmO8aiOCqgsKz6v5Kf BPG8yl0MSux40EmAhYDKF7wVPMhF0t1aJwa/XSISvotChUag1i/5VR8qo+JMKTltvxHuVwsumx+4 QbjXAF2goou3oovS8vN+uWjVZG47RpFQYmqNo9JY1e9JHzt9OYpKDnoF7LI3o6qDlQFM+PFFCQwQ R3R5xxx12Ezs4Q2nw0iqOk9rBCFcjgX05jOLnIQYA/XCVVUFqj+LQqGPKiyGanoFMt4MNCTaGeNO rhTzNtpx3zFXt9o/t6N7Y43B/ooVIOsqzb2x3WI5PSLIRKnl2S68HATHAAAAAAAA --=-lBQzFsK0B7j1nNmLRgp1--