[SYSDM] Prevent my german-teacher from getting a heart-attack by that misplaced comma.
[SMSS] Prevent myself from getting a heart-attack by that superfluous dot.
Furthermore this addendum serves the purpose of actually linking both PRs
and their 4 previous commits to the actual JIRA ticket CORE-18754, which wasn't
the case before:
0.4.15-dev-5392-g 04b2d35f5b
0.4.15-dev-5391-g a8e06d92e8
0.4.15-dev-5390-g a4274ad548
0.4.15-dev-5389-g 5dc43c0f32
What we have:
- Maximum number of pagefiles: 16
- Minimum pagefile size: 256 pages (1 MB when page size = 4096 bytes)
- Maximum pagefile size:
* 32-bit platforms: (1024 * 1024 - 1) pages (~ 4095 MB)
* x86 with PAE support: same size as for AMD x64
* x64 platform: (4 * 1024 * 1024 * 1024 - 1) pages (~ 16 TB)
* IA64 platform: (8 * 1024 * 1024 * 1024 - 1) pages (~ 32 TB)
Those are the values as supported and verified by the NT kernel.
Now, user-mode programs (including SMSS.EXE) have different opinions
on these, namely, they consider estimates directly in MB, respectively:
4095 MB, (16 * 1024 * 1024) MB, and (32 * 1024 * 1024) MB
(verified on Win2k3 and Win7 32 and 64 bits).
Also here, the minimum pagefile size is set to 2 MB.
Starting Windows 8+ (and 10), those values change slightly, and are
still not fully synchronized between NTOS:MM and SMSS. Finally, while
(x86 PAE and) AMD64 and ARM64 seem to share the maximum pagefile
size limit, 32-bit ARMv7 appears to use different limits than regular
x86 (2 GB instead of 4).
Please keep those values as they are for NT compatibility!
See the following references:
https://www.geoffchappell.com/studies/windows/km/ntoskrnl/api/mm/modwrite/create.htmhttps://techcommunity.microsoft.com/t5/ask-the-performance-team/what-is-the-page-file-for-anyway/ba-p/372608
+ Manual extraction of the values from different NT 6.2,6.3,10 builds.
[SMSS] Fill out in particular the x86-specific case for PAE.
[NTOS:MM] Some cleanup in the NtCreatePagingFile() code, namely:
- Clarify some comments;
- Validate the lower and upper bounds of the Minimum and Maximum sizes
(based on Windows behaviour as explained by Geoff + manual tests).
- Open the pagefile in case-insensitive;
- Simplify the loop that finds an existing matching pagefile;
- Simplify some failure exit paths;
- Add a "Missing validation steps TODO" comment block explaining the
existing code-hole.
This fixes starting the Windows 2000 POSIX subsystem in ReactOS.
- The CreateSession pointer was initialized against the SbApiMsg variable, but
it was the other SbApiMsg2 that was being initialized and sent through LPC.
- Do not overwrite the MuSessionId (Terminal Services session ID) variable with
the generated environment subsystem session ID from SmpAllocateSessionId().
- Actually initialize the SbApiMsg ApiNumber for the CreateSession LPC call.
(dll\win32\kernel32\client\proc.c:3690) Retrying with: POSIX /P C:\ReactOS\system32\posix\ls.exe /C ls
Breakpoint 1 hit
csrsrv!CsrSbApiRequestThread+0x64:
001b:1000ac34 837dfc00 cmp dword ptr [ebp-4],0
kd> ??ReceiveMsg
struct _SB_API_MSG
+0x000 h : _PORT_MESSAGE
+0x018 ConnectionInfo : _SB_CONNECTION_INFO
+0x018 ApiNumber : 0xcccccccc (No matching name)
+0x01c ReturnValue : 0n0
+0x020 u : <unnamed-tag>
kd> p
...
(base\system\smss\smsubsys.c:393) SMSS: SmpLoadSubSystem - NtRequestWaitReplyPort Failed with Status c0000002 for sessionid 2
...
<Retrying>
...
(base\system\smss\smsubsys.c:393) SMSS: SmpLoadSubSystem - NtRequestWaitReplyPort Failed with Status c0000002 for sessionid 3
All those bugs could have been avoided *IF*, rather than (badly) duplicating
its code, the existing SmpSbCreateSession() function had been used instead.
- "Not sure these field mean what I think they do -- but clear them" ... ◔_◔
Those fields are related to the debug client interface (DbgUi) and session
in case the subsystem being started is going to be debugged. These have
nothing to do with the MuSessionId. Clarify this in the SB_CREATE_SESSION_MSG
structure and in the SmpSbCreateSession() function.
The SubSystemNames array didn't correlate with the possible values of
SubSystemType (e.g. index 4 was "Posix" whereas Posix is type 7; Posix
and OS/2 entries were inverted; Windows CUI subsystem (type 3) was
mapped to "Posix"), and the array dereferencing was out of bounds if the
SubSystemType of the image happened to be larger than 8.
I know (strings extraction from debug build of Windows' SMSS.EXE) that
they use that same old'n'broken array. Perhaps a leftover from very old
times (NT 3.1 betas) where the PE format was under work and the
subsystem numbers didn't have their definitive values... (This has
already happened with the NT PDK v1.196 from September 1991.)
Loosely based on the deprecated ReactOS-specific SmExecuteProgram().
On server-side, we lookup into the list of deferred subsystems that
has been initialized at init time.
Dedicated to Justin Miller (The_DarkFire) work on reviving the
POSIX subsystem!
- Not all the wcscpy() / swprintf() calls have been converted to their
string-safe equivalents. Instead I used the string-safe functions only
for places where strings of unknown length were copied into fixed-size
internal buffers. On the contrary, for known-fixed-length strings being
copied or numbers being converted to string representations in large
enough buffers, I kept the original function calls.
- Verify the registry data that has been returned by NtQueryValueKey():
* When expecting (not multi) strings, check whether the data type is
either REG_SZ or REG_EXPAND_SZ.
* When expecting DWORD values, check whether the data type is
REG_DWORD and whether the data length is (greater or) equal to
sizeof(ULONG).
64 MB used to be enough to at least finish 2nd stage, but that does not
always seem to be the case anymore. Leaving this little free space does
not make for a good user experience either way.
256 MB is still not much, but at least provides slightly more usability.
The PROCESS_DEVICEMAP_INFORMATION union has 2 fields, one is a handle, the other one is a structure of 36 bytes (independent of architecture). The handle forces 64 bit alignment on 64 bit builds, making the structure 4 bytes bigger than on 32 bit builds. The site is checked in NtQueryInformationProcess (case ProcessDeviceMap). The expected size on x64 is the size of the Query structure without alignment. autocheck correctly passes the site of the Query union member, while smss passes the full size of PROCESS_DEVICEMAP_INFORMATION. Packing the structure is not an option, since it is defined in public headers without packing. Using the original headers sizeof(PROCESS_DEVICEMAP_INFORMATION) is 0x28, sizeof(PROCESS_DEVICEMAP_INFORMATION::Query) is 0x24.
