From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from e31.co.us.ibm.com (e31.co.us.ibm.com [32.97.110.149]) (using TLSv1 with cipher DHE-RSA-AES256-SHA (256/256 bits)) (Client CN "e31.co.us.ibm.com", Issuer "Equifax" (verified OK)) by ozlabs.org (Postfix) with ESMTP id C34E9679F0 for ; Fri, 23 Jun 2006 16:36:39 +1000 (EST) Received: from d03relay04.boulder.ibm.com (d03relay04.boulder.ibm.com [9.17.195.106]) by e31.co.us.ibm.com (8.12.11.20060308/8.12.11) with ESMTP id k5N6aYru020127 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=FAIL) for ; Fri, 23 Jun 2006 02:36:34 -0400 Received: from d03av02.boulder.ibm.com (d03av02.boulder.ibm.com [9.17.195.168]) by d03relay04.boulder.ibm.com (8.13.6/NCO/VER7.0) with ESMTP id k5N6akDG091440 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=NO) for ; Fri, 23 Jun 2006 00:36:46 -0600 Received: from d03av02.boulder.ibm.com (loopback [127.0.0.1]) by d03av02.boulder.ibm.com (8.12.11.20060308/8.13.3) with ESMTP id k5N6aXiC000858 for ; Fri, 23 Jun 2006 00:36:33 -0600 Message-ID: <449B8B9E.2050802@us.ibm.com> Date: Thu, 22 Jun 2006 23:35:10 -0700 From: Haren Myneni MIME-Version: 1.0 To: Paul Mackerras , Olof Johansson Subject: [PATCH] powerpc: kdump: Reserve the existing TCE mappings left by the first kernel Content-Type: multipart/mixed; boundary="------------000401070901050505010604" Cc: linuxppc-dev@ozlabs.org List-Id: Linux on PowerPC Developers Mail List List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , This is a multi-part message in MIME format. --------------000401070901050505010604 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Paul, Olof, tce_union got removed in powerpc.git tree. Hence reposting the patch created against this tree. This patch is implemented based on Olof's suggestion of reserving TCE entries and use only available ones in the kdump boot and tested on P5, P4, JS20 and JS21 machines. If you are OK with this patch, please send it to upstream. Thanks Haren During kdump boot, noticed some machines checkstop on dma protection fault for ongoing DMA left in the first kernel. Instead of initializing TCE entries in iommu_init() for the kdump boot, this patch fixes this issue by walking through the each TCE table and checks whether the entries are in use by the first kernel. If so, reserve those entries by setting the corresponding bit in tbl->it_map such that these entries will not be available for the kdump boot. However it could be possible that all TCE entries might be used up due to the driver bug that does continuous mapping. My observation is around 1700 TCE entries are used on some systems (Ex: P4) at some point of time during kdump boot and saving dump (either write into the disk or sending to remote machine). Hence, this patch will make sure that minimum of 2048 entries will be available such that kdump boot could be successful in some cases. Signed-off-by: Haren Myneni --------------000401070901050505010604 Content-Type: text/x-patch; name="ppc64-kdump-checkstop-DMA-fault-fix.patch" Content-Transfer-Encoding: 7bit Content-Disposition: inline; filename="ppc64-kdump-checkstop-DMA-fault-fix.patch" diff -Naurp powerpc.git.orig/arch/powerpc/kernel/iommu.c powerpc.git/arch/powerpc/kernel/iommu.c --- powerpc.git.orig/arch/powerpc/kernel/iommu.c 2006-06-19 00:54:37.000000000 -0500 +++ powerpc.git/arch/powerpc/kernel/iommu.c 2006-06-19 00:52:31.000000000 -0500 @@ -38,6 +38,7 @@ #include #include #include +#include #define DBG(...) @@ -440,8 +441,37 @@ struct iommu_table *iommu_init_table(str tbl->it_largehint = tbl->it_halfpoint; spin_lock_init(&tbl->it_lock); +#ifdef CONFIG_CRASH_DUMP + if (ppc_md.tce_get) { + unsigned long index, tceval; + unsigned long tcecount = 0; + + /* + * Reserve the existing mappings left by the first kernel. + */ + for (index = 0; index < tbl->it_size; index++) { + tceval = ppc_md.tce_get(tbl, index + tbl->it_offset); + /* + * Freed TCE entry contains 0x7fffffffffffffff on JS20 + */ + if (tceval && (tceval != 0x7fffffffffffffffUL)) { + __set_bit(index, tbl->it_map); + tcecount++; + } + } + if ((tbl->it_size - tcecount) < KDUMP_MIN_TCE_ENTRIES) { + printk(KERN_WARNING "TCE table is full; "); + printk(KERN_WARNING "freeing %d entries for the kdump boot\n", + KDUMP_MIN_TCE_ENTRIES); + for (index = tbl->it_size - KDUMP_MIN_TCE_ENTRIES; + index < tbl->it_size; index++) + __clear_bit(index, tbl->it_map); + } + } +#else /* Clear the hardware table in case firmware left allocations in it */ ppc_md.tce_free(tbl, tbl->it_offset, tbl->it_size); +#endif if (!welcomed) { printk(KERN_INFO "IOMMU table initialized, virtual merging %s\n", diff -Naurp powerpc.git.orig/arch/powerpc/platforms/pseries/iommu.c powerpc.git/arch/powerpc/platforms/pseries/iommu.c --- powerpc.git.orig/arch/powerpc/platforms/pseries/iommu.c 2006-06-19 00:54:27.000000000 -0500 +++ powerpc.git/arch/powerpc/platforms/pseries/iommu.c 2006-06-19 00:52:44.000000000 -0500 @@ -92,6 +92,15 @@ static void tce_free_pSeries(struct iomm *(tcep++) = 0; } +static unsigned long tce_get_pseries(struct iommu_table *tbl, long index) +{ + u64 *tcep; + + index <<= TCE_PAGE_FACTOR; + tcep = ((u64 *)tbl->it_base) + index; + + return *tcep; +} static void tce_build_pSeriesLP(struct iommu_table *tbl, long tcenum, long npages, unsigned long uaddr, @@ -235,6 +244,25 @@ static void tce_freemulti_pSeriesLP(stru } } +static unsigned long tce_get_pSeriesLP(struct iommu_table *tbl, long tcenum) +{ + u64 rc; + unsigned long tce_ret; + + tcenum <<= TCE_PAGE_FACTOR; + rc = plpar_tce_get((u64)tbl->it_index, (u64)tcenum << 12, &tce_ret); + + if (rc && printk_ratelimit()) { + printk("tce_get_pSeriesLP: plpar_tce_get failed. rc=%ld\n", + rc); + printk("\tindex = 0x%lx\n", (u64)tbl->it_index); + printk("\ttcenum = 0x%lx\n", (u64)tcenum); + show_stack(current, (unsigned long *)__get_SP()); + } + + return tce_ret; +} + static void iommu_table_setparms(struct pci_controller *phb, struct device_node *dn, struct iommu_table *tbl) @@ -254,7 +282,10 @@ static void iommu_table_setparms(struct } tbl->it_base = (unsigned long)__va(*basep); + +#ifndef CONFIG_CRASH_DUMP memset((void *)tbl->it_base, 0, *sizep); +#endif tbl->it_busno = phb->bus->number; @@ -564,11 +595,13 @@ void iommu_init_early_pSeries(void) ppc_md.tce_build = tce_build_pSeriesLP; ppc_md.tce_free = tce_free_pSeriesLP; } + ppc_md.tce_get = tce_get_pSeriesLP; ppc_md.iommu_bus_setup = iommu_bus_setup_pSeriesLP; ppc_md.iommu_dev_setup = iommu_dev_setup_pSeriesLP; } else { ppc_md.tce_build = tce_build_pSeries; ppc_md.tce_free = tce_free_pSeries; + ppc_md.tce_get = tce_get_pseries; ppc_md.iommu_bus_setup = iommu_bus_setup_pSeries; ppc_md.iommu_dev_setup = iommu_dev_setup_pSeries; } diff -Naurp powerpc.git.orig/include/asm-powerpc/kdump.h powerpc.git/include/asm-powerpc/kdump.h --- powerpc.git.orig/include/asm-powerpc/kdump.h 2006-06-19 00:55:11.000000000 -0500 +++ powerpc.git/include/asm-powerpc/kdump.h 2006-06-22 02:48:15.000000000 -0500 @@ -15,6 +15,8 @@ #define KDUMP_TRAMPOLINE_START 0x0100 #define KDUMP_TRAMPOLINE_END 0x3000 +#define KDUMP_MIN_TCE_ENTRIES 2048 + #else /* !CONFIG_CRASH_DUMP */ #define PHYSICAL_START 0x0 diff -Naurp powerpc.git.orig/include/asm-powerpc/machdep.h powerpc.git/include/asm-powerpc/machdep.h --- powerpc.git.orig/include/asm-powerpc/machdep.h 2006-06-19 00:54:58.000000000 -0500 +++ powerpc.git/include/asm-powerpc/machdep.h 2006-06-19 00:53:05.000000000 -0500 @@ -82,6 +82,8 @@ struct machdep_calls { void (*tce_free)(struct iommu_table *tbl, long index, long npages); + unsigned long (*tce_get)(struct iommu_table *tbl, + long index); void (*tce_flush)(struct iommu_table *tbl); void (*iommu_dev_setup)(struct pci_dev *dev); void (*iommu_bus_setup)(struct pci_bus *bus); --------------000401070901050505010604--