Reject MODE SELECT Format Device pages whose length differs from the 24-byte page returned by MODE SENSE. Use a complete valid page in the truncation regression and cover both invalid page lengths and transfer lengths exceeding the buffer.
Validate that the declared MODE SELECT parameter list fits in the transfer buffer and contains a complete command header. Validate the block-descriptor list against the remaining parameter-list length before skipping it.
Reject truncated mode-page headers and page bodies before reading them. Report ILLEGAL REQUEST / PARAMETER LIST LENGTH ERROR for truncated headers, block descriptors, and mode pages, as required by SCSI-2 8.2.8; retain INVALID FIELD IN PARAMETER LIST for complete but invalid fields.
Update the hard-disk and magneto-optical fixtures to send complete Format Device pages with their actual MODE SELECT parameter-list lengths. Add regression coverage for truncated headers, descriptors, page headers, and page bodies.
* Revert "Don't ResizeCache on sector change if no filename is defined (#1438)"
This reverts commit dd9a3296d4.
* Revert "Add ModeSense page 0x25 (DEC special function control page) (#1412)"
This reverts commit 1121b8d9d6.
* Revert "DiskCache needs a size"
This reverts commit 7cc8df271c.
* Revert "Honor sector size change via ModeSelect6 in scsicd (#1406)"
This reverts commit b7f65d33e2.
* Revert "Multiple fixes for ModeSelect (#1405)"
This reverts commit ad5eae93e7.
* Allow 'empty' ModeSelect6
tl;dr
Treat a computed length of 0 as `has_valid_page_code`.
Details:
The SRM console (aka 'BIOS') of DEC Alpha sends an empty
ModeSelect6 with the following data:
~~~
ModeSelect6, CDB $151000000c00
~~~
That makes 12 byte(s) as follows
~~~
0 1 2 3 4 5 6 7 8 9 10 11
00 00 00 08 00 00 00 00 00 00 02 00
~~~
decoding it (accoring to [1], Section 8.3.3, Table 94) gives us
Mode Data Length 0
Medium Type 0
Device-specific 0
Block desc len 8
Density Code 0
Number of blks 0
Reserved 0
Block length 512
`scsi_command_util::ModeSelect` computes
~~~
offset = 4 + buf[3];
~~~
giving 12 and
~~~
length -= offset;
~~~
giving 0.
Thus it never enters the `while` loop and `has_valid_page_code` stays
`false`, raising an error.
[1] [Small Computer System Interface - 2 rev 10L.pdf](https://dn790004.ca.archive.org/0/items/SCSISpecificationDocumentsSCSIDocuments/Small%20Computer%20System%20Interface%20-%202%20rev%2010L.pdf)
Signed-off-by: Klaus Kämpf <kkaempf@gmail.com>
* Allow ModeSelect with page code 1
OpenVMS Alpha (the operating system, not the SRM BIOS) uses
ModeSelect6 with a page code of 1.
The semantics are unknown, just accepting it works for me.
Signed-off-by: Klaus Kämpf <kkaempf@gmail.com>
* Fix page length computation in ModeSelect
tl;dr
The 'skip to next ModeSelect page' computation was off-by-one, either not
taking the page code itself into account or missing the fact that the
page length is given as `n - 1`.
Fix:
Add 1 to the computed length.
Details:
OpenVMS Alpha sends a ModeSelect6 as follows
~~~
command:
ModeSelect6, CDB $151000001900
payload:
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
00 00 00 08 00 00 00 00 00 00 02 00 01 0a 24 00 00 00 00 00 00 00 00 00 00
~~~
This translates to (accoring to [1], Section 8.3.3)
~~~
Mode Data Length 0
Medium Type 0
Device-specific 0
Block desc len 8
~~~
with the following offset / length computation _before_ the `while` loop
~~~
offset = 12
length = 13
~~~
The first payload section is
~~~
4 5 6 7 8 9 10 11
00 00 00 00 00 00 02 00
~~~
translating to
~~~
Density Code 0
Number of blks 0
Reserved 0
Block length 0x200 512
~~~
Then follows a pagecode 1 as
~~~
12 13 14 15 16 17 18 19 20 21 22 23 24
01 0a 24 00 00 00 00 00 00 00 00 00 00
~~~
translating to
~~~~
Page code 1
Page length -1 10
Mode parameters 24 00 00 00 00 00 00 00 00 00 00
~~~
computing (inside the `while` loop, as `// Advance to the next page`)
~~~
size = 10 + 2 = 12
~~~
followed by new `offset` and `length` values
~~~
offset = 25
length = 1
~~~
So it stays in the `while` loop (and has a larger-than-buffer `offset`
value)
Signed-off-by: Klaus Kämpf <kkaempf@gmail.com>
* Rebrand project to PiSCSI
- rascsi ->piscsi
- rasctl -> scsictl
- rasdump -> scsidump
- ras* -> piscsi* (rasutil -> piscsi_util, etc.)
* Refined the formatting and wording of the app startup banner
* Kept some references to rascsi and rasctl where backwards compatibility is concerned
* Point to the new github repo URL
Co-authored-by: nucleogenic <nr@nucleogenic.com>
Co-authored-by: Uwe Seimet <Uwe.Seimet@seimet.de>
* Support for ID and LUN parameter for the -L option in rascsi
* Added DeviceLogger class
* Removed dupiicate code
* Fixed SonarQube issues
* Added unit tests, improved code sharing
* Fixed regression (#979)
* Using lambdas instead of member function pointers simplifies the command dispatching and reduces the code volume
* Removed duplicate error handling
* Fix for issue #956
* Unit test updates
* Resolved SonarQube issues
* Unit test updates
* Lambda syntax cleanup
* Use new-style casts
* Use std::none_of when saving the cache
* Use to_integer instead of casts
* Use accessors for getting CDB data
* Made ctrl_t private
* Improved encapsulation
* Replaced pointers by references
* Removed all remaining occurrences of DWORD and BYTE, making os.h obsolete
- Moved C++ code to cpp/ from src/raspberrypi
- Related updates to Makefile, easyinstall.sh, and the github build rules
- Removed the native X68k C code in src/x68k from the repo
Daniel Markstedt
Klaus Kämpf
Uwe Seimet
