From mboxrd@z Thu Jan 1 00:00:00 1970 X-Spam-Checker-Version: SpamAssassin 3.4.4 (2020-01-24) on polar.synack.me X-Spam-Level: X-Spam-Status: No, score=-0.5 required=5.0 tests=BAYES_05 autolearn=ham autolearn_force=no version=3.4.4 X-Google-Thread: 103376,bc599f275917bfe9,start X-Google-Attributes: gid103376,public X-Google-Language: ENGLISH,ASCII-7-bit Path: g2news1.google.com!news1.google.com!news.glorb.com!border1.nntp.dca.giganews.com!border2.nntp.dca.giganews.com!nntp.giganews.com!wn14feed!worldnet.att.net!attbi_s04.POSTED!53ab2750!not-for-mail From: tmoran@acm.org Newsgroups: comp.lang.ada Subject: dynamic allocation X-Newsreader: Tom's custom newsreader Message-ID: <_eMXc.192913$8_6.100631@attbi_s04> NNTP-Posting-Host: 24.6.132.82 X-Complaints-To: abuse@comcast.net X-Trace: attbi_s04 1093636154 24.6.132.82 (Fri, 27 Aug 2004 19:49:14 GMT) NNTP-Posting-Date: Fri, 27 Aug 2004 19:49:14 GMT Organization: Comcast Online Date: Fri, 27 Aug 2004 19:49:14 GMT Xref: g2news1.google.com comp.lang.ada:3098 Date: 2004-08-27T19:49:14+00:00 List-Id: "during testing, an exception was made" >From ACM Tech News: >"NASA: DOS Glitch Nearly Killed Mars Rover" >Extreme Tech (08/23/04); Hachman, Mark > >NASA scientist Robert Denise said at this week's Hot Chips conference that >the real cause of a glitch on the Mars Spirit rover early this year was >not corruption in the flash memory, but rather an embedded DOS file system >that grew out of control. An undisclosed software vendor had required the >flash memory that stored the directory structure in RAM, which was only >half the size of the flash memory and eventually ran out of space. In >addition, there was only 128 MB of DRAM compared to 256 MB of flash >memory, virtually guaranteeing an eventual crash. The problem with the >file system was that the DOS directory structure was stored as a file that >did not reduce in size even when files were deleted from the directory >tree, since special characters indicated where new data could come in. >While that aspect alone was not dangerous, the Wind River embedded OS used >a piece of third-party software that required the mirroring of flash >memory in RAM. Denise admitted that during testing, an exception was made >for the dynamic allocation feature, which is usually not allowed in >embedded systems. When the system finally generated a memory allocation >command to a nonexistent memory address, it crashed. The error did not >manifest itself until Jan. 21, when Spirit fell into a continuous reboot >cycle and could not transmit data through its antenna. After five days, >NASA Jet Propulsion Laboratory's Flight Software Development Team solved >the problem by using system RAM as simulated flash and then isolating the >error. After flash memory was erased, the NASA team installed a >file-system monitoring utility that tracked memory as a consumable resource.