kernel.git/fs/proc/task_mmu.c, branch linux-2.6.28.y

pagemap: fix 32-bit pagemap regression

2008-12-10T16:01:53Z

The large pages fix from bcf8039ed45 broke 32-bit pagemap by pulling the pagemap entry code out into a function with the wrong return type. Pagemap entries are 64 bits on all systems and unsigned long is only 32 bits on 32-bit systems. Signed-off-by: Matt Mackall Reported-by: Doug Graham Cc: Alexey Dobriyan Cc: Dave Hansen Cc: [2.6.26.x, 2.6.27.x] Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds

proc: fix vma display mismatch between /proc/pid/{maps,smaps}

2008-10-23T09:21:29Z

Commit 4752c369789250eafcd7813e11c8fb689235b0d2 aka "maps4: simplify interdependence of maps and smaps" broke /proc/pid/smaps, causing it to display some vmas twice and other vmas not at all. For example: grep .- /proc/1/smaps >/tmp/smaps; diff /proc/1/maps /tmp/smaps 1 25d24 2 < 7fd7e23aa000-7fd7e23ac000 rw-p 7fd7e23aa000 00:00 0 3 28a28 4 > ffffffffff600000-ffffffffff601000 r-xp 00000000 00:00 0 [vsyscall] The bug has something to do with setting m->version before all the seq_printf's have been performed. show_map was doing this correctly, but show_smap was doing this in the middle of its seq_printf sequence. This patch arranges things so that the setting of m->version in show_smap is also done at the end of its seq_printf sequence. Testing: in addition to the above grep test, for each process I summed up the 'Rss' fields of /proc/pid/smaps and compared that to the 'VmRSS' field of /proc/pid/status. All matched except for Xorg (which has a /dev/mem mapping which Rss accounts for but VmRSS does not). This result gives us some confidence that neither /proc/pid/maps nor /proc/pid/smaps are any longer skipping or double-counting vmas. Signed-off-by: Joe Korty Cc: Matt Mackall Signed-off-by: Andrew Morton Signed-off-by: Alexey Dobriyan

proc: remove kernel.maps_protect

2008-10-10T00:24:51Z

After commit 831830b5a2b5d413407adf380ef62fe17d6fcbf2 aka "restrict reading from /proc//maps to those who share ->mm or can ptrace" sysctl stopped being relevant because commit moved security checks from ->show time to ->start time (mm_for_maps()). Signed-off-by: Alexey Dobriyan Acked-by: Kees Cook

/proc/self/maps doesn't display the real file offset

2008-08-20T22:40:30Z

This addresses http://bugzilla.kernel.org/show_bug.cgi?id=11318 In function show_map (file: fs/proc/task_mmu.c), if vma->vm_pgoff > 2^20 than (vma->vm_pgoff << PAGE_SIZE) is greater than 2^32 (with PAGE_SIZE equal to 4096 (i.e. 2^12). The next seq_printf use an unsigned long for the conversion of (vma->vm_pgoff << PAGE_SIZE), as a result the offset value displayed in /proc/self/maps is truncated if the page offset is greater than 2^20. A test that shows this issue: #define _GNU_SOURCE #include #include #include #include #include #include #include #include #define PAGE_SIZE (getpagesize()) #if __i386__ # define U64_STR "%llx" #elif __x86_64 # define U64_STR "%lx" #else # error "Architecture Unsupported" #endif int main(int argc, char *argv[]) { int fd; char *addr; off64_t offset = 0x10000000; char *filename = "/dev/zero"; fd = open(filename, O_RDONLY); if (fd < 0) { perror("open"); return 1; } offset *= 0x10; printf("offset = " U64_STR "\n", offset); addr = (char*)mmap64(NULL, PAGE_SIZE, PROT_READ, MAP_PRIVATE, fd, offset); if ((void*)addr == MAP_FAILED) { perror("mmap64"); return 1; } { FILE *fmaps; char *line = NULL; size_t len = 0; ssize_t read; size_t filename_len = strlen(filename); fmaps = fopen("/proc/self/maps", "r"); if (!fmaps) { perror("fopen"); return 1; } while ((read = getline(&line, &len, fmaps)) != -1) { if ((read > filename_len + 1) && (strncmp(&line[read - filename_len - 1], filename, filename_len) == 0)) printf("%s", line); } if (line) free(line); fclose(fmaps); } close(fd); return 0; } [akpm@linux-foundation.org: coding-style fixes] Signed-off-by: Clement Calmels Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds

proc: fix /proc/*/pagemap some more

2008-07-22T16:59:41Z

struct pagemap_walk was placed on stack, some hooks are initialized, the rest (->pgd_entry, ->pud_entry, ->pte_entry) are valid but junk. Reported-by: Eric Sesterhenn Signed-off-by: Alexey Dobriyan Cc: "Vegard Nossum" Cc: [2.6.25.x, 2.6.26.x] Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds

Security: split proc ptrace checking into read vs. attach

2008-07-14T05:01:47Z

Enable security modules to distinguish reading of process state via proc from full ptrace access by renaming ptrace_may_attach to ptrace_may_access and adding a mode argument indicating whether only read access or full attach access is requested. This allows security modules to permit access to reading process state without granting full ptrace access. The base DAC/capability checking remains unchanged. Read access to /proc/pid/mem continues to apply a full ptrace attach check since check_mem_permission() already requires the current task to already be ptracing the target. The other ptrace checks within proc for elements like environ, maps, and fds are changed to pass the read mode instead of attach. In the SELinux case, we model such reading of process state as a reading of a proc file labeled with the target process' label. This enables SELinux policy to permit such reading of process state without permitting control or manipulation of the target process, as there are a number of cases where programs probe for such information via proc but do not need to be able to control the target (e.g. procps, lsof, PolicyKit, ConsoleKit). At present we have to choose between allowing full ptrace in policy (more permissive than required/desired) or breaking functionality (or in some cases just silencing the denials via dontaudit rules but this can hide genuine attacks). This version of the patch incorporates comments from Casey Schaufler (change/replace existing ptrace_may_attach interface, pass access mode), and Chris Wright (provide greater consistency in the checking). Note that like their predecessors __ptrace_may_attach and ptrace_may_attach, the __ptrace_may_access and ptrace_may_access interfaces use different return value conventions from each other (0 or -errno vs. 1 or 0). I retained this difference to avoid any changes to the caller logic but made the difference clearer by changing the latter interface to return a bool rather than an int and by adding a comment about it to ptrace.h for any future callers. Signed-off-by: Stephen Smalley Acked-by: Chris Wright Signed-off-by: James Morris

Fix pagemap_read() use of struct mm_walk

2008-07-05T20:13:44Z

Fix some issues in pagemap_read noted by Alexey: - initialize pagemap_walk.mm to "mm" , so the code starts working as advertised - initialize ->private to "&pm" so it wouldn't immediately oops in pagemap_pte_hole() - unstatic struct pagemap_walk, so two threads won't fsckup each other (including those started by root, including flipping ->mm when you don't have permissions) - pagemap_read() contains two calls to ptrace_may_attach(), second one looks unneeded. - avoid possible kmalloc(0) and integer wraparound. Cc: Alexey Dobriyan Cc: Matt Mackall Signed-off-by: Andrew Morton [ Personally, I'd just remove the functionality entirely - Linus ] Signed-off-by: Linus Torvalds

Fix clear_refs_write() use of struct mm_walk

2008-07-05T20:07:56Z

Don't use a static entry, so as to prevent races during concurrent use of this function. Reported-by: Alexey Dobriyan Cc: Matt Mackall Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds

pagemap: fix large pages in pagemap

2008-06-13T01:05:41Z

We were walking right into huge page areas in the pagemap walker, and calling the pmds pmd_bad() and clearing them. That leaked huge pages. Bad. This patch at least works around that for now. It ignores huge pages in the pagemap walker for the time being, and won't leak those pages. Signed-off-by: Dave Hansen Acked-by: Matt Mackall Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds

pagemap: pass mm into pagewalkers

2008-06-13T01:05:41Z

We need this at least for huge page detection for now, because powerpc needs the vm_area_struct to be able to determine whether a virtual address is referring to a huge page (its pmd_huge() doesn't work). It might also come in handy for some of the other users. Signed-off-by: Dave Hansen Acked-by: Matt Mackall Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds