runtime crash

[lukacsarpad]

The attached code (ACM Algorithm 003: Finding roots of polynomials by the Bairstow-Hitchcock method) crashes on linux (ubuntu 16.04) with:
*** Error in `./001_bairstow2': free(): invalid next size (fast): 0x0000000000f8c0b0 ***
======= Backtrace: =========
/lib/x86_64-linux-gnu/libc.so.6(+0x777e5)[0x7f97f2ecf7e5]
/lib/x86_64-linux-gnu/libc.so.6(+0x8037a)[0x7f97f2ed837a]
/lib/x86_64-linux-gnu/libc.so.6(cfree+0x4c)[0x7f97f2edc53c]
./001_bairstow2[0x402010]
./001_bairstow2[0x401815]
./001_bairstow2[0x400c78]
/lib/x86_64-linux-gnu/libc.so.6(__libc_start_main+0xf0)[0x7f97f2e78830]
./001_bairstow2[0x400e49]
======= Memory map: ========
00400000-00405000 r-xp 00000000 09:00 7869045 /home/arpi/programming/a60/toms/001_bairstow2
00605000-00606000 r--p 00005000 09:00 7869045 /home/arpi/programming/a60/toms/001_bairstow2
00606000-00607000 rw-p 00006000 09:00 7869045 /home/arpi/programming/a60/toms/001_bairstow2
00f8c000-00fad000 rw-p 00000000 00:00 0 [heap]
7f97ec000000-7f97ec021000 rw-p 00000000 00:00 0
7f97ec021000-7f97f0000000 ---p 00000000 00:00 0
7f97f2c42000-7f97f2c58000 r-xp 00000000 08:21 5247481 /lib/x86_64-linux-gnu/libgcc_s.so.1
7f97f2c58000-7f97f2e57000 ---p 00016000 08:21 5247481 /lib/x86_64-linux-gnu/libgcc_s.so.1
7f97f2e57000-7f97f2e58000 rw-p 00015000 08:21 5247481 /lib/x86_64-linux-gnu/libgcc_s.so.1
7f97f2e58000-7f97f3018000 r-xp 00000000 08:21 5254184 /lib/x86_64-linux-gnu/libc-2.23.so
7f97f3018000-7f97f3218000 ---p 001c0000 08:21 5254184 /lib/x86_64-linux-gnu/libc-2.23.so
7f97f3218000-7f97f321c000 r--p 001c0000 08:21 5254184 /lib/x86_64-linux-gnu/libc-2.23.so
7f97f321c000-7f97f321e000 rw-p 001c4000 08:21 5254184 /lib/x86_64-linux-gnu/libc-2.23.so
7f97f321e000-7f97f3222000 rw-p 00000000 00:00 0
7f97f3222000-7f97f332a000 r-xp 00000000 08:21 5244227 /lib/x86_64-linux-gnu/libm-2.23.so
7f97f332a000-7f97f3529000 ---p 00108000 08:21 5244227 /lib/x86_64-linux-gnu/libm-2.23.so
7f97f3529000-7f97f352a000 r--p 00107000 08:21 5244227 /lib/x86_64-linux-gnu/libm-2.23.so
7f97f352a000-7f97f352b000 rw-p 00108000 08:21 5244227 /lib/x86_64-linux-gnu/libm-2.23.so
7f97f352b000-7f97f3551000 r-xp 00000000 08:21 5254182 /lib/x86_64-linux-gnu/ld-2.23.so
7f97f371f000-7f97f3723000 rw-p 00000000 00:00 0
7f97f374f000-7f97f3750000 rw-p 00000000 00:00 0
7f97f3750000-7f97f3751000 r--p 00025000 08:21 5254182 /lib/x86_64-linux-gnu/ld-2.23.so
7f97f3751000-7f97f3752000 rw-p 00026000 08:21 5254182 /lib/x86_64-linux-gnu/ld-2.23.so
7f97f3752000-7f97f3753000 rw-p 00000000 00:00 0
7ffd774af000-7ffd774d0000 rw-p 00000000 00:00 0 [stack]
7ffd775a5000-7ffd775a8000 r--p 00000000 00:00 0 [vvar]
7ffd775a8000-7ffd775aa000 r-xp 00000000 00:00 0 [vdso]
ffffffffff600000-ffffffffff601000 r-xp 00000000 00:00 0 [vsyscall]
Aborted (core dumped)

The same code (with fabs replaced by abs and using # instead of e for scientific form constants) runs with MARST and generates the output provided in the certification of the algorithm.
003_bairstow2.txt

[JvanKatwijk]

[jan@lenovo examples]$ ./003_bairstow2 Roots found by Bairstow's method. i ex nat x y 1 2 -1 -0.9706389700 1.005807589 2 1 -1 2.470638970 4.640533161 Roots found by Bairstow's method. i ex nat x y 1 1 -1 0.5000000000 0.8660254037 2 1 -1 -1.000000000 1.000000000 3 1 -1 1.500000000 1.322875655 Program end. If these are the correct output then the error is repaired. Simple error, had nothing to do with the 32 -> 64 bit transfer, but an incorrect counter Op do 27 dec. 2018 om 09:09 schreef lukacsarpad <notifications@github.com>:

…

The attached code (ACM Algorithm 003: Finding roots of polynomials by the Bairstow-Hitchcock method) crashes on linux (ubuntu 16.04) with: *** Error in `./001_bairstow2': free(): invalid next size (fast): 0x0000000000f8c0b0 *** ======= Backtrace: ========= /lib/x86_64-linux-gnu/libc.so.6(+0x777e5)[0x7f97f2ecf7e5] /lib/x86_64-linux-gnu/libc.so.6(+0x8037a)[0x7f97f2ed837a] /lib/x86_64-linux-gnu/libc.so.6(cfree+0x4c)[0x7f97f2edc53c] ./001_bairstow2[0x402010] ./001_bairstow2[0x401815] ./001_bairstow2[0x400c78] /lib/x86_64-linux-gnu/libc.so.6(__libc_start_main+0xf0)[0x7f97f2e78830] ./001_bairstow2[0x400e49] ======= Memory map: ======== 00400000-00405000 r-xp 00000000 09:00 7869045 /home/arpi/programming/a60/toms/001_bairstow2 00605000-00606000 r--p 00005000 09:00 7869045 /home/arpi/programming/a60/toms/001_bairstow2 00606000-00607000 rw-p 00006000 09:00 7869045 /home/arpi/programming/a60/toms/001_bairstow2 00f8c000-00fad000 rw-p 00000000 00:00 0 [heap] 7f97ec000000-7f97ec021000 rw-p 00000000 00:00 0 7f97ec021000-7f97f0000000 ---p 00000000 00:00 0 7f97f2c42000-7f97f2c58000 r-xp 00000000 08:21 5247481 /lib/x86_64-linux-gnu/libgcc_s.so.1 7f97f2c58000-7f97f2e57000 ---p 00016000 08:21 5247481 /lib/x86_64-linux-gnu/libgcc_s.so.1 7f97f2e57000-7f97f2e58000 rw-p 00015000 08:21 5247481 /lib/x86_64-linux-gnu/libgcc_s.so.1 7f97f2e58000-7f97f3018000 r-xp 00000000 08:21 5254184 /lib/x86_64-linux-gnu/libc-2.23.so 7f97f3018000-7f97f3218000 ---p 001c0000 08:21 5254184 /lib/x86_64-linux-gnu/libc-2.23.so 7f97f3218000-7f97f321c000 r--p 001c0000 08:21 5254184 /lib/x86_64-linux-gnu/libc-2.23.so 7f97f321c000-7f97f321e000 rw-p 001c4000 08:21 5254184 /lib/x86_64-linux-gnu/libc-2.23.so 7f97f321e000-7f97f3222000 rw-p 00000000 00:00 0 7f97f3222000-7f97f332a000 r-xp 00000000 08:21 5244227 /lib/x86_64-linux-gnu/libm-2.23.so 7f97f332a000-7f97f3529000 ---p 00108000 08:21 5244227 /lib/x86_64-linux-gnu/libm-2.23.so 7f97f3529000-7f97f352a000 r--p 00107000 08:21 5244227 /lib/x86_64-linux-gnu/libm-2.23.so 7f97f352a000-7f97f352b000 rw-p 00108000 08:21 5244227 /lib/x86_64-linux-gnu/libm-2.23.so 7f97f352b000-7f97f3551000 r-xp 00000000 08:21 5254182 /lib/x86_64-linux-gnu/ld-2.23.so 7f97f371f000-7f97f3723000 rw-p 00000000 00:00 0 7f97f374f000-7f97f3750000 rw-p 00000000 00:00 0 7f97f3750000-7f97f3751000 r--p 00025000 08:21 5254182 /lib/x86_64-linux-gnu/ld-2.23.so 7f97f3751000-7f97f3752000 rw-p 00026000 08:21 5254182 /lib/x86_64-linux-gnu/ld-2.23.so 7f97f3752000-7f97f3753000 rw-p 00000000 00:00 0 7ffd774af000-7ffd774d0000 rw-p 00000000 00:00 0 [stack] 7ffd775a5000-7ffd775a8000 r--p 00000000 00:00 0 [vvar] 7ffd775a8000-7ffd775aa000 r-xp 00000000 00:00 0 [vdso] ffffffffff600000-ffffffffff601000 r-xp 00000000 00:00 0 [vsyscall] Aborted (core dumped) The same code (with fabs replaced by abs and using # instead of e for scientific form constants) runs with MARST and generates the output provided in the certification of the algorithm. 003_bairstow2.txt <https://github.com/JvanKatwijk/algol-60-compiler/files/2711886/003_bairstow2.txt> — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#2>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AITzwEvIk_oQAuyKMwEL4oltkz5LkOOCks5u9IAdgaJpZM4Zid5S> .

-- Jan van Katwijk +31 (0)15 3698980 +31 (0) 628260355

[lukacsarpad]

Yes, this seems correct, thank you.

Also, I wanted to ask you if you are sure that fabs() shall be used for floating point numbers. MARST seems to accept abs(), and the original TOMS algorithms use abs() too.

[JvanKatwijk]

I really have to consult the RR for that, technically is is not an issue. Op do 27 dec. 2018 om 14:44 schreef lukacsarpad <notifications@github.com>:

…

Yes, this seems correct, thank you. Also, I wanted to ask you if you are sure that fabs() shall be used for floating point numbers. MARST seems to accept abs(), and the original TOMS algorithms use abs() too. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#2 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AITzwE5kkhiDSwFdjcEtc42T-BDbvou-ks5u9M7ZgaJpZM4Zid5S> .

-- Jan van Katwijk +31 (0)15 3698980 +31 (0) 628260355

[JvanKatwijk]

fabs changed in abs, seems to work Op do 27 dec. 2018 om 15:02 schreef jan van katwijk <j.vankatwijk@gmail.com

…

: I really have to consult the RR for that, technically is is not an issue. Op do 27 dec. 2018 om 14:44 schreef lukacsarpad ***@***.*** >: > Yes, this seems correct, thank you. > > Also, I wanted to ask you if you are sure that fabs() shall be used for > floating point numbers. MARST seems to accept abs(), and the original TOMS > algorithms use abs() too. > > — > You are receiving this because you commented. > Reply to this email directly, view it on GitHub > <#2 (comment)>, > or mute the thread > <https://github.com/notifications/unsubscribe-auth/AITzwE5kkhiDSwFdjcEtc42T-BDbvou-ks5u9M7ZgaJpZM4Zid5S> > . > -- Jan van Katwijk +31 (0)15 3698980 +31 (0) 628260355

-- Jan van Katwijk +31 (0)15 3698980 +31 (0) 628260355

[lukacsarpad]

Somehow I cannot reproduce the correct results with the new version. I get
Roots found by Bairstow's method.
i ex nat x y
1 1 -1 -1.012085580 1.008068111
2 1 -1 2.512085580 4.671048047
Roots found by Bairstow's method.
i ex nat x y
1 3 -1 0.5073488146 0.8601905079
2 1 1 -2.926409128e-3 -632542.6365
3 1 -1 316271.8123 547798.2775
Program end.

[JvanKatwijk]

Thanks The problem was caused by the mapping of the algol "abs" onto the c "abs", the latter operating on int's The solution was to create a "fabs" in the runtime, and an "abs" in the prelude that just calls the fabs The results I get now, are the correct results Op zo 6 jan. 2019 om 17:53 schreef lukacsarpad <notifications@github.com>:

…

Somehow I cannot reproduce the correct results with the new version. I get Roots found by Bairstow's method. i ex nat x y 1 1 -1 -1.012085580 1.008068111 2 1 -1 2.512085580 4.671048047 Roots found by Bairstow's method. i ex nat x y 1 3 -1 0.5073488146 0.8601905079 2 1 1 -2.926409128e-3 -632542.6365 3 1 -1 316271.8123 547798.2775 Program end. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#2 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AITzwChhrB8QmBoQQyO_TbWnCOHoU3QRks5vAioXgaJpZM4Zid5S> .

-- Jan van Katwijk +31 (0)15 3698980 +31 (0) 628260355