crash patch c3413456599161cabc4e910a0ae91dfe5eec3c21 (xen: Add support for dom0 with Linux kernel 3.19 and newer) from Daniel made crash utility support xen dom0 vmcores after linux kernel commit 054954eb051f35e74b75a566a96fe756015352c8 (xen: switch to linear virtual mapped sparse p2m list). This patch can be deemed as a subsequent and make this utility support Xen PV domU dumpfiles again. Basically speaking, readmem() can't be used to read xen_p2m_addr associate memory directly during m2p translation. It introduces infinite recursion. Following call sequence shows the scenario, it comes from a section of backtrace with only kvaddr, machine addr and mfn left as parameter: module_init() /* The first readmem() from module_init(). */ readmem(addr=0xffffffffa02fe4a0) /* readmem() needs physical address, so calls kvtop(). */ kvtop(kvaddr=0xffffffffa02fe4a0) x86_64_kvtop(kvaddr=ffffffffa02fe4a0) /* Calculate physical address by traversing page tables. */ x86_64_kvtop_xen_wpt(kvaddr=0xffffffffa02fe4a0) /* * x86_64_kvtop_xen_wpt() is going to traverse the page table to * get the physical address for 0xffffffffa02fe4a0. So, at first it * is needed to translate the pgd from machine address to physical * address. So invoke xen_m2p() here to do the translation. 0x58687f000 * is the pgd machine address in x86_64_kvtop_xen_wpt() and is needed * to be translated to its physical address. */ xen_m2p(machine=0x58687f000) __xen_m2p(machine=0x58687f000, mfn=0x58687f) /* * __xen_m2p() is going to search mfn 0x58687f in p2m VMA which starts * at VMA 0xffffc900001cf000. It compares every mfn stored in it with * 0x58687f. Once it's proved 0x58687f is one mfn in the p2m, its offset * will be used to calculate the pfn. * * readmem() is invoked by __xen_m2p() to read the page from VMA * 0xffffc900001cf000 here. */ readmem(addr=0xffffc900001cf000) /* * readmem() needs physical address of 0xffffc900001cf000 to make the * reading done. So it invokes kvtop() to get the physical address. */ kvtop(kvaddr=0xffffc900001cf000) x86_64_kvtop(kvaddr=0xffffc900001cf000) /* It needs to calculate physical address by traversing page tables. */ x86_64_kvtop_xen_wpt(kvaddr=0xffffc900001cf000) /* * 0x581b7e000 is the machine address of pgd need to be translated here. * The mfn is calculated in this way at x86_64_kvtop_xen_wpt(): * * pml4 = ((ulong *)machdep->machspec->pml4) + pml4_index(kvaddr); * pgd_paddr = (*pml4) & PHYSICAL_PAGE_MASK; * * The kvaddr 0xffffc900001cf000 here is quite different from the one * above, so the machine address of pgd is not the same one. And this * pgd is the one we use to access the VMA of p2m table. */ xen_m2p(machine=0x581b7e000) __xen_m2p(machine=0x581b7e000, mfn=0x581b7e) /* * Looking for mfn 0x581b7e in the range of p2m page which starts at * VMA 0xffffc900001f5000. */ readmem(addr=0xffffc900001f5000) /* Need physical address of VMA 0xffffc900001f5000 as same reason above. */ kvtop(kvaddr=0xffffc900001f5000) x86_64_kvtop(kvaddr=0xffffc900001f5000) /* Need to traverse page tables to calculate physical address for it. */ x86_64_kvtop_xen_wpt(kvaddr=0xffffc900001f5000) /* * Unfortunately, machine address 0x581b7e000 have to be translated again. * Endless loop starts from here. */ xen_m2p(machine=0x581b7e000) __xen_m2p(machine=0x581b7e000, mfn=0x581b7e) readmem(addr=0xffffc900001f5000) Fortunately, PV domU p2m mapping is also stored at xd->xfd + xch_index_offset and organized as struct xen_dumpcore_p2m. We have a chance to read the p2m stuff directly from there, and then we avoid the loop above. So, this patch implements a special reading function read_xc_p2m() to extract the mfns from xd->xfd + xch_index_offset. This function does not need to read mfns from p2m VMA like readmem() does, so, we avoid the endless loop introduced by the address translation. Signed-off-by: Honglei Wang <honglei.wang(a)oracle.com&gt; Reviewed-by: Daniel Kiper <daniel.kiper(a)oracle.com&gt; --- kernel.c | 151 +++++++++++++++++++++++++++++++++++++++++++++++++++++++------ xendump.c | 2 +- xendump.h | 2 + 3 files changed, 140 insertions(+), 15 deletions(-) diff --git a/kernel.c b/kernel.c index 395736c..7a5ce64 100644 --- a/kernel.c +++ b/kernel.c @@ -22,6 +22,7 @@ #include <libgen.h> #include <ctype.h> #include <stdbool.h> +#include "xendump.h" static void do_module_cmd(ulong, char *, ulong, char *, char *); static void show_module_taint(void); @@ -67,6 +68,9 @@ static ulong __xen_m2p(ulonglong, ulong); static ulong __xen_pvops_m2p_l2(ulonglong, ulong); static ulong __xen_pvops_m2p_l3(ulonglong, ulong); static ulong __xen_pvops_m2p_hyper(ulonglong, ulong); +static ulong __xen_pvops_m2p_domU(ulonglong, ulong); +static int read_xc_p2m(ulonglong addr, void *buffer, long size); +static void read_p2m(ulong cache_index, int memtype, void *buffer); static int search_mapping_page(ulong, ulong *, ulong *, ulong *); static void read_in_kernel_config_err(int, char *); static void BUG_bytes_init(void); @@ -181,10 +185,7 @@ kernel_init() &kt->pvops_xen.p2m_mid_missing); get_symbol_data("p2m_missing", sizeof(ulong), &kt->pvops_xen.p2m_missing); - } else if (symbol_exists("xen_p2m_addr")) { - if (!XEN_CORE_DUMPFILE()) - error(FATAL, "p2m array in new format is unreadable."); - } else { + } else if (!symbol_exists("xen_p2m_addr")) { kt->pvops_xen.p2m_top_entries = get_array_length("p2m_top", NULL, 0); kt->pvops_xen.p2m_top = symbol_value("p2m_top"); kt->pvops_xen.p2m_missing = symbol_value("p2m_missing"); @@ -9305,13 +9306,7 @@ __xen_m2p(ulonglong machine, ulong mfn) if (memtype == PHYSADDR) pc->curcmd_flags |= XEN_MACHINE_ADDR; - if (!readmem(kt->p2m_mapping_cache[c].mapping, memtype, - mp, PAGESIZE(), "phys_to_machine_mapping page (cached)", - RETURN_ON_ERROR)) - error(FATAL, "cannot access " - "phys_to_machine_mapping page\n"); - else - kt->last_mapping_read = kt->p2m_mapping_cache[c].mapping; + read_p2m(c, memtype, mp); if (memtype == PHYSADDR) pc->curcmd_flags &= ~XEN_MACHINE_ADDR; @@ -9349,9 +9344,12 @@ __xen_m2p(ulonglong machine, ulong mfn) */ if (symbol_exists("p2m_mid_missing")) pfn = __xen_pvops_m2p_l3(machine, mfn); - else if (symbol_exists("xen_p2m_addr")) - pfn = __xen_pvops_m2p_hyper(machine, mfn); - else + else if (symbol_exists("xen_p2m_addr")) { + if (XEN_CORE_DUMPFILE()) + pfn = __xen_pvops_m2p_hyper(machine, mfn); + else + pfn = __xen_pvops_m2p_domU(machine, mfn); + } else pfn = __xen_pvops_m2p_l2(machine, mfn); if (pfn != XEN_MFN_NOT_FOUND) @@ -9559,6 +9557,131 @@ __xen_pvops_m2p_hyper(ulonglong machine, ulong mfn) return XEN_MFN_NOT_FOUND; } +static void read_p2m(ulong cache_index, int memtype, void *buffer) +{ + /* + * Use special read function for PV domain p2m reading. + * See the comments of read_xc_p2m(). + */ + if (symbol_exists("xen_p2m_addr") && !XEN_CORE_DUMPFILE()) { + if (!read_xc_p2m(kt->p2m_mapping_cache[cache_index].mapping, + buffer, PAGESIZE())) + error(FATAL, "cannot access phys_to_machine_mapping page\n"); + } else if (!readmem(kt->p2m_mapping_cache[cache_index].mapping, memtype, + buffer, PAGESIZE(), "phys_to_machine_mapping page (cached)", + RETURN_ON_ERROR)) + error(FATAL, "cannot access phys_to_machine_mapping page\n"); + + kt->last_mapping_read = kt->p2m_mapping_cache[cache_index].mapping; +} + +/* + * PV domain p2m mapping info is stored in xd->xfd at xch_index_offset. It + * is organized as struct xen_dumpcore_p2m and the pfns are progressively + * increased by 1 from 0. + * + * This is a special p2m reading function for xen PV domain vmcores after + * kernel commit 054954eb051f35e74b75a566a96fe756015352c8 (xen: switch + * to linear virtual mapped sparse p2m list). It is invoked for reading + * p2m associate stuff by read_p2m(). + */ +static int read_xc_p2m(ulonglong addr, void *buffer, long size) +{ + ulong i, new_p2m_buf_size; + off_t offset; + struct xen_dumpcore_p2m *new_p2m_buf; + static struct xen_dumpcore_p2m *p2m_buf; + static ulong p2m_buf_size = 0; + + if (size <= 0) { + if ((CRASHDEBUG(1) && !STREQ(pc->curcmd, "search")) || + CRASHDEBUG(2)) + error(INFO, "invalid size request: %ld\n", size); + return FALSE; + } + + /* + * We extract xen_dumpcore_p2m.gmfn and copy them into the + * buffer. So, we need temporary p2m_buf whose size is + * (size * (sizeof(struct xen_dumpcore_p2m) / sizeof(ulong))) + * to put xen_dumpcore_p2m structures read from xd->xfd. + */ + new_p2m_buf_size = size * (sizeof(struct xen_dumpcore_p2m) / sizeof(ulong)); + + if (p2m_buf_size != new_p2m_buf_size) { + p2m_buf_size = new_p2m_buf_size; + + new_p2m_buf = realloc(p2m_buf, p2m_buf_size); + if (new_p2m_buf == NULL) { + free(p2m_buf); + error(FATAL, "cannot realloc p2m buffer\n"); + } + p2m_buf = new_p2m_buf; + } + + offset = addr * (sizeof(struct xen_dumpcore_p2m) / sizeof(ulong)); + offset += xd->xc_core.header.xch_index_offset; + + if (lseek(xd->xfd, offset, SEEK_SET) == -1) + error(FATAL, + "cannot lseek to xch_index_offset offset 0x%lx\n", offset); + if (read(xd->xfd, (void*)p2m_buf, p2m_buf_size) != p2m_buf_size) + error(FATAL, + "cannot read from xch_index_offset offset 0x%lx\n", offset); + + for (i = 0; i < size / sizeof(ulong); i++) + *((ulong *)buffer + i) = p2m_buf[i].gmfn; + + return TRUE; +} + +static ulong +__xen_pvops_m2p_domU(ulonglong machine, ulong mfn) +{ + ulong c, end, i, mapping, p, pfn, start; + + /* + * xch_nr_pages is the number of pages of p2m mapping. It is composed + * of struct xen_dumpcore_p2m. The stuff we want to copy into the mapping + * page is mfn whose type is unsigned long. + * So actual number of p2m pages should be: + * + * xch_nr_pages / (sizeof(struct xen_dumpcore_p2m) / sizeof(ulong)) + */ + for (p = 0; + p < xd->xc_core.header.xch_nr_pages / + (sizeof(struct xen_dumpcore_p2m) / sizeof(ulong)); + ++p) { + + mapping = p * PAGESIZE(); + + if (mapping != kt->last_mapping_read) { + if (!read_xc_p2m(mapping, (void *)kt->m2p_page, PAGESIZE())) + error(FATAL, "cannot read the last mapping page\n"); + kt->last_mapping_read = mapping; + } + kt->p2m_pages_searched++; + + if (search_mapping_page(mfn, &i, &start, &end)) { + pfn = p * XEN_PFNS_PER_PAGE + i; + c = kt->p2m_cache_index; + if (CRASHDEBUG (1)) + console("mfn: %lx (%llx) i: %ld pfn: %lx (%llx)\n", + mfn, machine, i, pfn, XEN_PFN_TO_PSEUDO(pfn)); + + kt->p2m_mapping_cache[c].start = start; + kt->p2m_mapping_cache[c].end = end; + kt->p2m_mapping_cache[c].mapping = mapping; + kt->p2m_mapping_cache[c].pfn = p * XEN_PFNS_PER_PAGE; + kt->p2m_cache_index = (c+1) % P2M_MAPPING_CACHE; + + return pfn; + } + } + + return XEN_MFN_NOT_FOUND; +} + /* * Search for an mfn in the current mapping page, and if found, * determine the range of contiguous mfns that it's contained diff --git a/xendump.c b/xendump.c index 8170b22..4bd59b5 100644 --- a/xendump.c +++ b/xendump.c @@ -19,7 +19,7 @@ #include "xendump.h" static struct xendump_data xendump_data = { 0 }; -static struct xendump_data *xd = &xendump_data; +struct xendump_data *xd = &xendump_data; static int xc_save_verify(char *); static int xc_core_verify(char *, char *); diff --git a/xendump.h b/xendump.h index 08d41b4..b7bae65 100644 --- a/xendump.h +++ b/xendump.h @@ -192,3 +192,5 @@ struct xen_dumpcore_p2m { uint64_t pfn; uint64_t gmfn; }; + +extern struct xendump_data *xd; -- 1.7.1