aix revision 258945
1258945SrobertoProblem with recent ANSI compilers 2258945Sroberto 3258945SrobertoOn some systems, including AIX, the compiler quits on the ntp_refclock.c 4258945Srobertofile when processing the refclock_report() routine. The problem, which 5258945Srobertois eithre a feature or a bug, has to do with an unwanted promotion of 6258945Srobertothe u_char argument to an int and a failure of the compiler to recognize 7258945Srobertothe preceding prototype. A workaround is to use ANSI syntax to delare 8258945Srobertothe arguments. Since ANSI compilers are not universally available, this 9258945Srobertosyntax can't be used in the stock distribution. 10258945Sroberto 11258945Sroberto(Message # 60: 2884 bytes, New) 12258945SrobertoDate: Sat, 19 Aug 1995 13:20:50 -0400 13258945SrobertoFrom: "R. Bernstein" <rocky@panix.com> 14258945SrobertoNewsgroups: comp.protocols.time.ntp 15258945Srobertoto: mills@udel.edu 16258945Srobertoreturn-receipt-to: rocky@panix.com 17258945SrobertoSubject: time and AIX 3.2.5 raw tty bug 18258945Sroberto 19258945SrobertoThis posting isn't strictly about NTP, any program that may stop the 20258945Srobertoclock or set the clock backwards is subject to the AIX 3.2.5 bug. 21258945Sroberto 22258945SrobertoOn AIX 3.2.5, there is a bug in the tty driver for a raw device which 23258945Srobertomay crash the box under certain conditions: basically a read() on a 24258945Srobertoraw tty in effect, a character was read but not as many as specified 25258945Srobertoby VMIN when a read timeout occurred. VTIME specifies the timeout. See 26258945Srobertothe AIX manual page on termios.h or that include file. for Information 27258945Srobertoon VMIN (or MIN) VTIME (or TIME). 28258945Sroberto 29258945SrobertoA remedy other than to not use raw tty's is to apply patch U435110. 30258945Sroberto 31258945SrobertoDetails of the problem report follow. 32258945Sroberto 33258945Sroberto> ABSTRACT: 34258945Sroberto> IX43779: TRAP IN PSX_TIMEO 35258945Sroberto> 36258945Sroberto> ORIGINATING DETAILS: 37258945Sroberto> Stacktrace shows: 38258945Sroberto> IAR: 01460214 posixdd:psx_timeo + 8bf4: ti 4,r12,0x0 39258945Sroberto> *LR: 014601a0 posixdd:psx_timeo + 8b80 40258945Sroberto> 00212c60: 014604f4 posixdd:psx_timer + 8ed4 41258945Sroberto> 00212cc0: 0144b74c ttydd:tty_do_offlevel + 4284 42258945Sroberto> 00212d20: 000216fc .i_offlevel + 8c 43258945Sroberto> 00212d70: 00021d78 .i_softint + c8 44258945Sroberto> 00001004: 00008714 .finish_interrupt + 80 45258945Sroberto> 46258945Sroberto> RESPONDER SUMMARY: 47258945Sroberto> AIX asserted in psx_timeo(). Reason for the assert was that 48258945Sroberto> the current time was behind psx_ctime. Since this state 49258945Sroberto> can occur when the current time is changed after a character 50258945Sroberto> is received but before the VTIME interbyte timer pops, we 51258945Sroberto> should not assert on this. 52258945Sroberto> 53258945Sroberto> RESPONDER CONCLUSION: 54258945Sroberto> Removed the requirement that current time > psx_ctime by 55258945Sroberto> adding a new L_ntimersub macro that is used instead of the 56258945Sroberto> ntimersub macro in time.h. Also added a test for (current 57258945Sroberto> time - psx_ctime) being negative, in that case we do not 58258945Sroberto> adjust the new timeout. 59258945Sroberto> 60258945Sroberto> Reported to Correct a PTF in Error: NO 61258945Sroberto> Reported as a Highly pervasive problem: NO 62258945Sroberto> 63258945Sroberto> PE Apar?: NoPE 64258945Sroberto> Hiper Apar?: NoHiper 65258945Sroberto> Status: CLOSED PER 66258945Sroberto> Component Name: AIX V3 FOR RS/6 67258945Sroberto> Version: 320 68258945Sroberto> Component ID: 575603001 69258945Sroberto> Submitted: 94/05/03 70258945Sroberto> Closed: 94/05/05 71258945Sroberto> ChangeTeam: TX2527 72258945Sroberto> 73258945Sroberto> APAR FIXED BY: U431696 U432151 U432844 U432870 U432979 74258945Sroberto> U433049 U433081 U433459 U433876 U433906 U434598 U434453 75258945Sroberto> U434672 U434737 U435110 76258945Sroberto 77