)]}' { "log": [ { "commit": "aaf230424204864e2833dcc1da23e2cb0b9f39cd", "tree": "a42042f5135aa63a780964bd053ae174211ab62f", "parents": [ "a261670aed2b5b77a20ce9b15fed57abeb126c0e" ], "author": { "name": "Yinghai Lu", "email": "Yinghai.Lu@Sun.COM", "time": "Wed Jan 30 13:33:09 2008 +0100" }, "committer": { "name": "Ingo Molnar", "email": "mingo@elte.hu", "time": "Wed Jan 30 13:33:09 2008 +0100" }, "message": "x86: disable the GART early, 64-bit\n\nFor K8 system: 4G RAM with memory hole remapping enabled, or more than\n4G RAM installed.\n\nwhen try to use kexec second kernel, and the first doesn\u0027t include\ngart_shutdown. the second kernel could have different aper position than\nthe first kernel. and second kernel could use that hole as RAM that is\nstill used by GART set by the first kernel. esp. when try to kexec\n2.6.24 with sparse mem enable from previous kernel (from RHEL 5 or SLES\n10). the new kernel will use aper by GART (set by first kernel) for\nvmemmap. and after new kernel setting one new GART. the position will be\nreal RAM. the _mapcount set is lost.\n\nBad page state in process \u0027swapper\u0027\npage:ffffe2000e600020 flags:0x0000000000000000 mapping:0000000000000000 mapcount:1 count:0\nTrying to fix it up, but a reboot is needed\nBacktrace:\nPid: 0, comm: swapper Not tainted 2.6.24-rc7-smp-gcdf71a10-dirty #13\n\nCall Trace:\n [\u003cffffffff8026401f\u003e] bad_page+0x63/0x8d\n [\u003cffffffff80264169\u003e] __free_pages_ok+0x7c/0x2a5\n [\u003cffffffff80ba75d1\u003e] free_all_bootmem_core+0xd0/0x198\n [\u003cffffffff80ba3a42\u003e] numa_free_all_bootmem+0x3b/0x76\n [\u003cffffffff80ba3461\u003e] mem_init+0x3b/0x152\n [\u003cffffffff80b959d3\u003e] start_kernel+0x236/0x2c2\n [\u003cffffffff80b9511a\u003e] _sinittext+0x11a/0x121\n\nand\n [ffffe2000e600000-ffffe2000e7fffff] PMD -\u003effff81001c200000 on node 0\nphys addr is : 0x1c200000\n\nRHEL 5.1 kernel -53 said:\nPCI-DMA: aperture base @ 1c000000 size 65536 KB\n\nnew kernel said:\nMapping aperture over 65536 KB of RAM @ 3c000000\n\nSo could try to disable that GART if possible.\n\nAccording to Ingo\n\n\u003e hm, i\u0027m wondering, instead of modifying the GART, why dont we simply\n\u003e _detect_ whatever GART settings we have inherited, and propagate that\n\u003e into our e820 maps? I.e. if there\u0027s inconsistency, then punch that out\n\u003e from the memory maps and just dont use that memory.\n\u003e\n\u003e that way it would not matter whether the GART settings came from a [old\n\u003e or crashing] Linux kernel that has not called gart_iommu_shutdown(), or\n\u003e whether it\u0027s a BIOS that has set up an aperture hole inconsistent with\n\u003e the memory map it passed. (or the memory map we _think_ i tried to pass\n\u003e us)\n\u003e\n\u003e it would also be more robust to only read and do a memory map quirk\n\u003e based on that, than actively trying to change the GART so early in the\n\u003e bootup. Later on we have to re-enable the GART _anyway_ and have to\n\u003e punch a hole for it.\n\u003e\n\u003e and as a bonus, we would have shored up our defenses against crappy\n\u003e BIOSes as well.\n\nadd e820 modification for gart inconsistent setting.\n\ngart_fix_e820\u003doff could be used to disable e820 fix.\n\nSigned-off-by: Yinghai Lu \u003cyinghai.lu@sun.com\u003e\nSigned-off-by: Ingo Molnar \u003cmingo@elte.hu\u003e\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\n" }, { "commit": "0440d4c00dadf8b97657c0864e4d54a088430de8", "tree": "4750cbab8af0b2b948519553ca19a724fdce8d24", "parents": [ "79da0874410c0af8958f0ee3a50aa5b67eecc640" ], "author": { "name": "Joerg Roedel", "email": "joerg.roedel@amd.com", "time": "Wed Oct 24 12:49:50 2007 +0200" }, "committer": { "name": "Thomas Gleixner", "email": "tglx@linutronix.de", "time": "Tue Oct 30 00:22:22 2007 +0100" }, "message": "x86 gart: rename symbols only used for the GART implementation\n\nThis patch renames the 4 symbols iommu_hole_init(), iommu_aperture,\niommu_aperture_allowed, iommu_aperture_disabled. All these symbols are only\nused for the GART implementation of IOMMUs.\n\nIt adds and additional gart_ prefix to them.\n\nSigned-off-by: Joerg Roedel \u003cjoerg.roedel@amd.com\u003e\nSigned-off-by: Ingo Molnar \u003cmingo@elte.hu\u003e\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\n" }, { "commit": "966396d3a05c8049fce5c81c49138e5ee1b05443", "tree": "11fff010a4aecd965bed52719d55d8fa4b40b055", "parents": [ "395624fcddd178de01a78aa88670a86ec919de77" ], "author": { "name": "Joerg Roedel", "email": "joerg.roedel@amd.com", "time": "Wed Oct 24 12:49:48 2007 +0200" }, "committer": { "name": "Thomas Gleixner", "email": "tglx@linutronix.de", "time": "Tue Oct 30 00:22:22 2007 +0100" }, "message": "x86 gart: rename CONFIG_IOMMU to CONFIG_GART_IOMMU\n\nThis patch renames the IOMMU config option to GART_IOMMU because in fact it\nmeans the GART and not general support for an IOMMU on x86.\n\nSigned-off-by: Joerg Roedel \u003cjoerg.roedel@amd.com\u003e\nAcked-by: Muli Ben-Yehuda \u003cmuli@il.ibm.com\u003e\nSigned-off-by: Ingo Molnar \u003cmingo@elte.hu\u003e\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\n" }, { "commit": "395624fcddd178de01a78aa88670a86ec919de77", "tree": "bfc66eafdbf5b220020a8d986fac0c6f255aeb51", "parents": [ "e1054b39b275340f7ba0308eca83c127b36e936e" ], "author": { "name": "Joerg Roedel", "email": "joerg.roedel@amd.com", "time": "Wed Oct 24 12:49:47 2007 +0200" }, "committer": { "name": "Thomas Gleixner", "email": "tglx@linutronix.de", "time": "Tue Oct 30 00:22:22 2007 +0100" }, "message": "x86 gart: rename iommu.h to gart.h\n\nThis patch renames the include file asm-x86/iommu.h to asm-x86/gart.h to make\nclear to which IOMMU implementation it belongs. The patch also adds \"GART\" to\nthe Kconfig line.\n\nSigned-off-by: Joerg Roedel \u003cjoerg.roedel@amd.com\u003e\nAcked-by: Muli Ben-Yehuda \u003cmuli@il.ibm.com\u003e\nSigned-off-by: Ingo Molnar \u003cmingo@elte.hu\u003e\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\n" } ] }