| 1 |
============================= |
| 2 |
No-MMU memory mapping support |
| 3 |
============================= |
| 4 |
|
| 5 |
The kernel has limited support for memory mapping under no-MMU conditions, such |
| 6 |
as are used in uClinux environments. From the userspace point of view, memory |
| 7 |
mapping is made use of in conjunction with the mmap() system call, the shmat() |
| 8 |
call and the execve() system call. From the kernel's point of view, execve() |
| 9 |
mapping is actually performed by the binfmt drivers, which call back into the |
| 10 |
mmap() routines to do the actual work. |
| 11 |
|
| 12 |
Memory mapping behaviour also involves the way fork(), vfork(), clone() and |
| 13 |
ptrace() work. Under uClinux there is no fork(), and clone() must be supplied |
| 14 |
the CLONE_VM flag. |
| 15 |
|
| 16 |
The behaviour is similar between the MMU and no-MMU cases, but not identical; |
| 17 |
and it's also much more restricted in the latter case: |
| 18 |
|
| 19 |
(#) Anonymous mapping, MAP_PRIVATE |
| 20 |
|
| 21 |
In the MMU case: VM regions backed by arbitrary pages; copy-on-write |
| 22 |
across fork. |
| 23 |
|
| 24 |
In the no-MMU case: VM regions backed by arbitrary contiguous runs of |
| 25 |
pages. |
| 26 |
|
| 27 |
(#) Anonymous mapping, MAP_SHARED |
| 28 |
|
| 29 |
These behave very much like private mappings, except that they're |
| 30 |
shared across fork() or clone() without CLONE_VM in the MMU case. Since |
| 31 |
the no-MMU case doesn't support these, behaviour is identical to |
| 32 |
MAP_PRIVATE there. |
| 33 |
|
| 34 |
(#) File, MAP_PRIVATE, PROT_READ / PROT_EXEC, !PROT_WRITE |
| 35 |
|
| 36 |
In the MMU case: VM regions backed by pages read from file; changes to |
| 37 |
the underlying file are reflected in the mapping; copied across fork. |
| 38 |
|
| 39 |
In the no-MMU case: |
| 40 |
|
| 41 |
- If one exists, the kernel will reuse an existing mapping to the |
| 42 |
same segment of the same file if that has compatible permissions, |
| 43 |
even if this was created by another process. |
| 44 |
|
| 45 |
- If possible, the file mapping will be directly on the backing device |
| 46 |
if the backing device has the NOMMU_MAP_DIRECT capability and |
| 47 |
appropriate mapping protection capabilities. Ramfs, romfs, cramfs |
| 48 |
and mtd might all permit this. |
| 49 |
|
| 50 |
- If the backing device can't or won't permit direct sharing, |
| 51 |
but does have the NOMMU_MAP_COPY capability, then a copy of the |
| 52 |
appropriate bit of the file will be read into a contiguous bit of |
| 53 |
memory and any extraneous space beyond the EOF will be cleared |
| 54 |
|
| 55 |
- Writes to the file do not affect the mapping; writes to the mapping |
| 56 |
are visible in other processes (no MMU protection), but should not |
| 57 |
happen. |
| 58 |
|
| 59 |
(#) File, MAP_PRIVATE, PROT_READ / PROT_EXEC, PROT_WRITE |
| 60 |
|
| 61 |
In the MMU case: like the non-PROT_WRITE case, except that the pages in |
| 62 |
question get copied before the write actually happens. From that point |
| 63 |
on writes to the file underneath that page no longer get reflected into |
| 64 |
the mapping's backing pages. The page is then backed by swap instead. |
| 65 |
|
| 66 |
In the no-MMU case: works much like the non-PROT_WRITE case, except |
| 67 |
that a copy is always taken and never shared. |
| 68 |
|
| 69 |
(#) Regular file / blockdev, MAP_SHARED, PROT_READ / PROT_EXEC / PROT_WRITE |
| 70 |
|
| 71 |
In the MMU case: VM regions backed by pages read from file; changes to |
| 72 |
pages written back to file; writes to file reflected into pages backing |
| 73 |
mapping; shared across fork. |
| 74 |
|
| 75 |
In the no-MMU case: not supported. |
| 76 |
|
| 77 |
(#) Memory backed regular file, MAP_SHARED, PROT_READ / PROT_EXEC / PROT_WRITE |
| 78 |
|
| 79 |
In the MMU case: As for ordinary regular files. |
| 80 |
|
| 81 |
In the no-MMU case: The filesystem providing the memory-backed file |
| 82 |
(such as ramfs or tmpfs) may choose to honour an open, truncate, mmap |
| 83 |
sequence by providing a contiguous sequence of pages to map. In that |
| 84 |
case, a shared-writable memory mapping will be possible. It will work |
| 85 |
as for the MMU case. If the filesystem does not provide any such |
| 86 |
support, then the mapping request will be denied. |
| 87 |
|
| 88 |
(#) Memory backed blockdev, MAP_SHARED, PROT_READ / PROT_EXEC / PROT_WRITE |
| 89 |
|
| 90 |
In the MMU case: As for ordinary regular files. |
| 91 |
|
| 92 |
In the no-MMU case: As for memory backed regular files, but the |
| 93 |
blockdev must be able to provide a contiguous run of pages without |
| 94 |
truncate being called. The ramdisk driver could do this if it allocated |
| 95 |
all its memory as a contiguous array upfront. |
| 96 |
|
| 97 |
(#) Memory backed chardev, MAP_SHARED, PROT_READ / PROT_EXEC / PROT_WRITE |
| 98 |
|
| 99 |
In the MMU case: As for ordinary regular files. |
| 100 |
|
| 101 |
In the no-MMU case: The character device driver may choose to honour |
| 102 |
the mmap() by providing direct access to the underlying device if it |
| 103 |
provides memory or quasi-memory that can be accessed directly. Examples |
| 104 |
of such are frame buffers and flash devices. If the driver does not |
| 105 |
provide any such support, then the mapping request will be denied. |
| 106 |
|
| 107 |
|
| 108 |
Further notes on no-MMU MMAP |
| 109 |
============================ |
| 110 |
|
| 111 |
(#) A request for a private mapping of a file may return a buffer that is not |
| 112 |
page-aligned. This is because XIP may take place, and the data may not be |
| 113 |
paged aligned in the backing store. |
| 114 |
|
| 115 |
(#) A request for an anonymous mapping will always be page aligned. If |
| 116 |
possible the size of the request should be a power of two otherwise some |
| 117 |
of the space may be wasted as the kernel must allocate a power-of-2 |
| 118 |
granule but will only discard the excess if appropriately configured as |
| 119 |
this has an effect on fragmentation. |
| 120 |
|
| 121 |
(#) The memory allocated by a request for an anonymous mapping will normally |
| 122 |
be cleared by the kernel before being returned in accordance with the |
| 123 |
Linux man pages (ver 2.22 or later). |
| 124 |
|
| 125 |
In the MMU case this can be achieved with reasonable performance as |
| 126 |
regions are backed by virtual pages, with the contents only being mapped |
| 127 |
to cleared physical pages when a write happens on that specific page |
| 128 |
(prior to which, the pages are effectively mapped to the global zero page |
| 129 |
from which reads can take place). This spreads out the time it takes to |
| 130 |
initialize the contents of a page - depending on the write-usage of the |
| 131 |
mapping. |
| 132 |
|
| 133 |
In the no-MMU case, however, anonymous mappings are backed by physical |
| 134 |
pages, and the entire map is cleared at allocation time. This can cause |
| 135 |
significant delays during a userspace malloc() as the C library does an |
| 136 |
anonymous mapping and the kernel then does a memset for the entire map. |
| 137 |
|
| 138 |
However, for memory that isn't required to be precleared - such as that |
| 139 |
returned by malloc() - mmap() can take a MAP_UNINITIALIZED flag to |
| 140 |
indicate to the kernel that it shouldn't bother clearing the memory before |
| 141 |
returning it. Note that CONFIG_MMAP_ALLOW_UNINITIALIZED must be enabled |
| 142 |
to permit this, otherwise the flag will be ignored. |
| 143 |
|
| 144 |
uClibc uses this to speed up malloc(), and the ELF-FDPIC binfmt uses this |
| 145 |
to allocate the brk and stack region. |
| 146 |
|
| 147 |
(#) A list of all the private copy and anonymous mappings on the system is |
| 148 |
visible through /proc/maps in no-MMU mode. |
| 149 |
|
| 150 |
(#) A list of all the mappings in use by a process is visible through |
| 151 |
/proc/<pid>/maps in no-MMU mode. |
| 152 |
|
| 153 |
(#) Supplying MAP_FIXED or a requesting a particular mapping address will |
| 154 |
result in an error. |
| 155 |
|
| 156 |
(#) Files mapped privately usually have to have a read method provided by the |
| 157 |
driver or filesystem so that the contents can be read into the memory |
| 158 |
allocated if mmap() chooses not to map the backing device directly. An |
| 159 |
error will result if they don't. This is most likely to be encountered |
| 160 |
with character device files, pipes, fifos and sockets. |
| 161 |
|
| 162 |
|
| 163 |
Interprocess shared memory |
| 164 |
========================== |
| 165 |
|
| 166 |
Both SYSV IPC SHM shared memory and POSIX shared memory is supported in NOMMU |
| 167 |
mode. The former through the usual mechanism, the latter through files created |
| 168 |
on ramfs or tmpfs mounts. |
| 169 |
|
| 170 |
|
| 171 |
Futexes |
| 172 |
======= |
| 173 |
|
| 174 |
Futexes are supported in NOMMU mode if the arch supports them. An error will |
| 175 |
be given if an address passed to the futex system call lies outside the |
| 176 |
mappings made by a process or if the mapping in which the address lies does not |
| 177 |
support futexes (such as an I/O chardev mapping). |
| 178 |
|
| 179 |
|
| 180 |
No-MMU mremap |
| 181 |
============= |
| 182 |
|
| 183 |
The mremap() function is partially supported. It may change the size of a |
| 184 |
mapping, and may move it [#]_ if MREMAP_MAYMOVE is specified and if the new size |
| 185 |
of the mapping exceeds the size of the slab object currently occupied by the |
| 186 |
memory to which the mapping refers, or if a smaller slab object could be used. |
| 187 |
|
| 188 |
MREMAP_FIXED is not supported, though it is ignored if there's no change of |
| 189 |
address and the object does not need to be moved. |
| 190 |
|
| 191 |
Shared mappings may not be moved. Shareable mappings may not be moved either, |
| 192 |
even if they are not currently shared. |
| 193 |
|
| 194 |
The mremap() function must be given an exact match for base address and size of |
| 195 |
a previously mapped object. It may not be used to create holes in existing |
| 196 |
mappings, move parts of existing mappings or resize parts of mappings. It must |
| 197 |
act on a complete mapping. |
| 198 |
|
| 199 |
.. [#] Not currently supported. |
| 200 |
|
| 201 |
|
| 202 |
Providing shareable character device support |
| 203 |
============================================ |
| 204 |
|
| 205 |
To provide shareable character device support, a driver must provide a |
| 206 |
file->f_op->get_unmapped_area() operation. The mmap() routines will call this |
| 207 |
to get a proposed address for the mapping. This may return an error if it |
| 208 |
doesn't wish to honour the mapping because it's too long, at a weird offset, |
| 209 |
under some unsupported combination of flags or whatever. |
| 210 |
|
| 211 |
The driver should also provide backing device information with capabilities set |
| 212 |
to indicate the permitted types of mapping on such devices. The default is |
| 213 |
assumed to be readable and writable, not executable, and only shareable |
| 214 |
directly (can't be copied). |
| 215 |
|
| 216 |
The file->f_op->mmap() operation will be called to actually inaugurate the |
| 217 |
mapping. It can be rejected at that point. Returning the ENOSYS error will |
| 218 |
cause the mapping to be copied instead if NOMMU_MAP_COPY is specified. |
| 219 |
|
| 220 |
The vm_ops->close() routine will be invoked when the last mapping on a chardev |
| 221 |
is removed. An existing mapping will be shared, partially or not, if possible |
| 222 |
without notifying the driver. |
| 223 |
|
| 224 |
It is permitted also for the file->f_op->get_unmapped_area() operation to |
| 225 |
return -ENOSYS. This will be taken to mean that this operation just doesn't |
| 226 |
want to handle it, despite the fact it's got an operation. For instance, it |
| 227 |
might try directing the call to a secondary driver which turns out not to |
| 228 |
implement it. Such is the case for the framebuffer driver which attempts to |
| 229 |
direct the call to the device-specific driver. Under such circumstances, the |
| 230 |
mapping request will be rejected if NOMMU_MAP_COPY is not specified, and a |
| 231 |
copy mapped otherwise. |
| 232 |
|
| 233 |
.. important:: |
| 234 |
|
| 235 |
Some types of device may present a different appearance to anyone |
| 236 |
looking at them in certain modes. Flash chips can be like this; for |
| 237 |
instance if they're in programming or erase mode, you might see the |
| 238 |
status reflected in the mapping, instead of the data. |
| 239 |
|
| 240 |
In such a case, care must be taken lest userspace see a shared or a |
| 241 |
private mapping showing such information when the driver is busy |
| 242 |
controlling the device. Remember especially: private executable |
| 243 |
mappings may still be mapped directly off the device under some |
| 244 |
circumstances! |
| 245 |
|
| 246 |
|
| 247 |
Providing shareable memory-backed file support |
| 248 |
============================================== |
| 249 |
|
| 250 |
Provision of shared mappings on memory backed files is similar to the provision |
| 251 |
of support for shared mapped character devices. The main difference is that the |
| 252 |
filesystem providing the service will probably allocate a contiguous collection |
| 253 |
of pages and permit mappings to be made on that. |
| 254 |
|
| 255 |
It is recommended that a truncate operation applied to such a file that |
| 256 |
increases the file size, if that file is empty, be taken as a request to gather |
| 257 |
enough pages to honour a mapping. This is required to support POSIX shared |
| 258 |
memory. |
| 259 |
|
| 260 |
Memory backed devices are indicated by the mapping's backing device info having |
| 261 |
the memory_backed flag set. |
| 262 |
|
| 263 |
|
| 264 |
Providing shareable block device support |
| 265 |
======================================== |
| 266 |
|
| 267 |
Provision of shared mappings on block device files is exactly the same as for |
| 268 |
character devices. If there isn't a real device underneath, then the driver |
| 269 |
should allocate sufficient contiguous memory to honour any supported mapping. |
| 270 |
|
| 271 |
|
| 272 |
Adjusting page trimming behaviour |
| 273 |
================================= |
| 274 |
|
| 275 |
NOMMU mmap automatically rounds up to the nearest power-of-2 number of pages |
| 276 |
when performing an allocation. This can have adverse effects on memory |
| 277 |
fragmentation, and as such, is left configurable. The default behaviour is to |
| 278 |
aggressively trim allocations and discard any excess pages back in to the page |
| 279 |
allocator. In order to retain finer-grained control over fragmentation, this |
| 280 |
behaviour can either be disabled completely, or bumped up to a higher page |
| 281 |
watermark where trimming begins. |
| 282 |
|
| 283 |
Page trimming behaviour is configurable via the sysctl ``vm.nr_trim_pages``. |