DISQUS

Ricardo Martins · 7 months ago

Arch Linux -- kernel 2.6.29.3, gcc 4.4.0
Intel Core2Duo 2.4 GHz (Macbook Pro 4,1)

scrypt failed to build with the following compiler errors:

scrypt-nosse.c: In function ‘scrypt’:
scrypt-nosse.c:298: error: ‘MAP_NOCORE’ undeclared (first use in this function)
scrypt-nosse.c:298: error: (Each undeclared identifier is reported only once
scrypt-nosse.c:298: error: for each function it appears in.)
make[1]: *** [scrypt-nosse.o] Error 1
make[1]: Leaving directory `/home/meqif/tmp/scrypt-1.1'
make: *** [all] Error 2

cperciva · 7 months ago

Oops. I coded the test for MAP_NOCORE backwards -- please try again with version 1.1.1.

Ricardo Martins · 7 months ago

Ah, now it compiled successfully. It also works fine with both a plain text file and a PDF (encryption and decryption). Thanks for fixing it!

Rory Arms · 7 months ago

Hi Colin. It fails to build on OSX 10.4.11/powerpc & 10.5.6/powerpc

First on 10.4, configure looks clean as far as I can tell, but the make step fails:

osse.Tpo -c -o scrypt-nosse.o scrypt-nosse.c
scrypt-nosse.c: In function ‘scrypt’:
scrypt-nosse.c:298: error: ‘MAP_NOCORE’ undeclared (first use in this function)
scrypt-nosse.c:298: error: (Each undeclared identifier is reported only once
scrypt-nosse.c:298: error: for each function it appears in.)
make[1]: *** [scrypt-nosse.o] Error 1
make: *** [all] Error 2

On 10.5.6/powerpc

I do see some errors in the configure output that you might find useful:

checking for a BSD-compatible install... /usr/bin/install -c
checking whether build environment is sane... yes
checking for a thread-safe mkdir -p... config.aux/install-sh -c -d
checking for gawk... no
checking for mawk... no
checking for nawk... no
checking for awk... awk
checking whether make sets $(MAKE)... yes
checking whether to enable maintainer-specific portions of Makefiles... no
checking for gcc... gcc
checking for C compiler default output file name... a.out
checking whether the C compiler works... yes
checking whether we are cross compiling... no
checking for suffix of executables...
checking for suffix of object files... o
checking whether we are using the GNU C compiler... yes
checking whether gcc accepts -g... yes
checking for gcc option to accept ISO C89... none needed
checking for style of include used by make... GNU
checking dependency style of gcc... gcc3
checking how to run the C preprocessor... gcc -E
checking for grep that handles long lines and -e... /usr/bin/grep
checking for egrep... /usr/bin/grep -E
checking for ANSI C header files... rm: conftest.dSYM/Contents/Resources/DWARF: Operation not permitted
rm: conftest.dSYM/Contents/Resources: Directory not empty
rm: conftest.dSYM/Contents: Directory not empty
rm: conftest.dSYM: Directory not empty
yes
checking for sys/types.h... yes
checking for sys/stat.h... yes
checking for stdlib.h... yes
checking for string.h... yes
checking for memory.h... yes
checking for strings.h... yes
checking for inttypes.h... yes
checking for stdint.h... yes
checking for unistd.h... yes
checking err.h usability... yes
checking err.h presence... yes
checking for err.h... yes
checking sys/endian.h usability... no
checking sys/endian.h presence... no
checking for sys/endian.h... no
checking sys/sysinfo.h usability... no
checking sys/sysinfo.h presence... no
checking for sys/sysinfo.h... no
checking for clock_gettime in -lrt... no
checking for clock_gettime... no
checking for sysinfo... no
checking for posix_memalign... no
checking for struct sysinfo... no
checking for struct sysinfo.mem_unit... no
configure: creating ./config.status
config.status: creating Makefile
config.status: creating config.h
config.status: executing depfiles commands

And the make failure:

scrypt-nosse.c: In function ‘scrypt’:
scrypt-nosse.c:298: error: ‘MAP_NOCORE’ undeclared (first use in this function)
scrypt-nosse.c:298: error: (Each undeclared identifier is reported only once
scrypt-nosse.c:298: error: for each function it appears in.)
make[1]: *** [scrypt-nosse.o] Error 1
make: *** [all] Error 2

ekarulf · 7 months ago

The same failure occurs on OS X 10.5.6 / Intel

cperciva · 7 months ago

I coded the test for MAP_NOCORE backwards -- please try again with version 1.1.1 (where I think I've fixed this).

Antonio Touriño · 7 months ago

Hi Colin,

I couldn't verify your scrypt-sigs-1.1.asc file. gpg complained it could not find the key ID 3DD61E72 on the public servers. Is there another place I should be looking for it?

Cheers,
Antonio

cperciva · 7 months ago

This is the tarsnap code signing GPG key, as mentioned at http://www.daemonology.net/blog/2009-01-17-tars...

Antonio Touriño · 7 months ago

Thanks for the pointer. scrypt has compiled and is working on MacOS 10.5.7 on an Intel iMac. I have only done light tests and it has worked correctly. I will report back if I find any bugs. Cheers.

da44en · 7 months ago

$ uname -a; make
SunOS keithmoon 5.10 Generic_127127-11 sun4v sparc SUNW,Sun-Fire-T200 Solaris
make all-am
gcc -DHAVE_CONFIG_H -I. -g -O2 -MT main.o -MD -MP -MF .deps/main.Tpo -c -o main.o main.c
mv -f .deps/main.Tpo .deps/main.Po
gcc -DHAVE_CONFIG_H -I. -g -O2 -MT scryptenc.o -MD -MP -MF .deps/scryptenc.Tpo -c -o scryptenc.o scryptenc.c
mv -f .deps/scryptenc.Tpo .deps/scryptenc.Po
gcc -DHAVE_CONFIG_H -I. -g -O2 -MT crypto_aesctr.o -MD -MP -MF .deps/crypto_aesctr.Tpo -c -o crypto_aesctr.o crypto_aesctr.c
mv -f .deps/crypto_aesctr.Tpo .deps/crypto_aesctr.Po
gcc -DHAVE_CONFIG_H -I. -g -O2 -MT sha256.o -MD -MP -MF .deps/sha256.Tpo -c -o sha256.o sha256.c
mv -f .deps/sha256.Tpo .deps/sha256.Po
gcc -DHAVE_CONFIG_H -I. -g -O2 -MT memlimit.o -MD -MP -MF .deps/memlimit.Tpo -c -o memlimit.o memlimit.c
memlimit.c: In function `memlimit_sysinfo':
memlimit.c:106: error: structure has no member named `totalram'
*** Error code 1
make: Fatal error: Command failed for target `memlimit.o'
Current working directory /home/darren/tmp/scrypt-1.1.1
*** Error code 1
make: Fatal error: Command failed for target `all'

cperciva · 7 months ago

Thanks! I should have added a check to make sure that the build was running on linux before assuming that any sysinfo function I detected was linux's sysinfo. If you edit config.h and delete the HAVE_SYSINFO line, does that fix the build?

da44en · 7 months ago

With that change, it compiles, but I get a runtime error:

$ ./scrypt enc config.h.in config.h.in.enc
Please enter passphrase:
Please confirm passphrase:
scrypt: Error reading clocks: Invalid argument

(Passphrase is foo.)

I get the same error when trying to decrypt a file encrypted by scrypt compiled on linux. I see that error is from main.c:230.

cperciva · 7 months ago

Thanks! Could you send your config.log and config.h files to me at cperciva@tarsnap.com so that I can look at them?

royce · 7 months ago

Your target demographic may not be running on a NetBSD 2.x system running on Alpha, but just in case ...

tycho@sdf$ uname -a
NetBSD sdf 2.1.0_STABLE NetBSD 2.1.0_STABLE (sdf) #0: Fri Mar 30 02:24:32 UTC 2007 root@ol:/var/sys/arch/alpha/compile/sdf alpha

tycho@sdf$ ./configure
checking for a BSD-compatible install... /usr/pkg/bin/ginstall -c
checking whether build environment is sane... yes
checking for a thread-safe mkdir -p... /usr/pkg/bin/gmkdir -p
checking for gawk... gawk
checking whether make sets $(MAKE)... yes
checking whether to enable maintainer-specific portions of Makefiles... no
checking for gcc... gcc
checking for C compiler default output file name... a.out
checking whether the C compiler works... yes
checking whether we are cross compiling... no
checking for suffix of executables...
checking for suffix of object files... o
checking whether we are using the GNU C compiler... yes
checking whether gcc accepts -g... yes
checking for gcc option to accept ISO C89... none needed
checking for style of include used by make... GNU
checking dependency style of gcc... gcc3
checking how to run the C preprocessor... gcc -E
checking for grep that handles long lines and -e... /usr/bin/grep
checking for egrep... /usr/bin/grep -E
checking for ANSI C header files... yes
checking for sys/types.h... yes
checking for sys/stat.h... yes
checking for stdlib.h... yes
checking for string.h... yes
checking for memory.h... yes
checking for strings.h... yes
checking for inttypes.h... yes
checking for stdint.h... yes
checking for unistd.h... yes
checking err.h usability... yes
checking err.h presence... yes
checking for err.h... yes
checking sys/endian.h usability... yes
checking sys/endian.h presence... yes
checking for sys/endian.h... yes
checking sys/sysinfo.h usability... no
checking sys/sysinfo.h presence... no
checking for sys/sysinfo.h... no
checking for clock_gettime in -lrt... yes
checking for clock_gettime... yes
checking for sysinfo... no
checking for posix_memalign... no
checking for struct sysinfo... no
checking for struct sysinfo.mem_unit... no
configure: creating ./config.status
config.status: creating Makefile
config.status: creating config.h
config.status: executing depfiles commands

tycho@sdf$ make
make all-am
gcc -DHAVE_CONFIG_H -I. -g -O2 -MT main.o -MD -MP -MF .deps/main.Tpo -c -o main.o main.c
mv -f .deps/main.Tpo .deps/main.Po
gcc -DHAVE_CONFIG_H -I. -g -O2 -MT scryptenc.o -MD -MP -MF .deps/scryptenc.Tpo -c -o scryptenc.o scryptenc.c
mv -f .deps/scryptenc.Tpo .deps/scryptenc.Po
gcc -DHAVE_CONFIG_H -I. -g -O2 -MT crypto_aesctr.o -MD -MP -MF .deps/crypto_aesctr.Tpo -c -o crypto_aesctr.o crypto_aesctr.c
mv -f .deps/crypto_aesctr.Tpo .deps/crypto_aesctr.Po
gcc -DHAVE_CONFIG_H -I. -g -O2 -MT sha256.o -MD -MP -MF .deps/sha256.Tpo -c -o sha256.o sha256.c
mv -f .deps/sha256.Tpo .deps/sha256.Po
gcc -DHAVE_CONFIG_H -I. -g -O2 -MT memlimit.o -MD -MP -MF .deps/memlimit.Tpo -c -o memlimit.o memlimit.c
memlimit.c: In function `memlimit_rlimit':
memlimit.c:138: error: `RLIMIT_AS' undeclared (first use in this function)
memlimit.c:138: error: (Each undeclared identifier is reported only once
memlimit.c:138: error: for each function it appears in.)
*** Error code 1

Stop.
make: stopped in /arpa/tz/t/tycho/src/scrypt-1.1.1
*** Error code 1

Stop.
make: stopped in /arpa/tz/t/tycho/src/scrypt-1.1.1

cperciva · 7 months ago

Thanks! It looks like NetBSD only added RLIMIT_AS recently, in spite of it being required by POSIX. Could you try wrapping the RLIMIT_AS check in memlimit.c with #ifdef RLIMIT_AS ... #endif and let me know if that fixes the build?

royce · 7 months ago

Another stop - some semi-duplicate lines removed for brevity:

gcc -g -O2 -o scrypt main.o scryptenc.o crypto_aesctr.o sha256.o memlimit.o warn.o scrypt_cpuperf.o scrypt-nosse.o -lcrypto -lrt

scryptenc.o(.text+0x718): In function `scryptenc_setup':
/arpa/tz/t/tycho/src/scrypt-1.1.1/scryptenc.c:207: undefined reference to `be32enc'

crypto_aesctr.o(.text+0x144): In function `crypto_aesctr_stream':
/arpa/tz/t/tycho/src/scrypt-1.1.1/crypto_aesctr.c:95: undefined reference to `be64enc'

sha256.o(.text+0x10c):/arpa/tz/t/tycho/src/scrypt-1.1.1/sha256.c:58: undefined reference to `be32dec'
sha256.o(.text+0x3228): In function `PBKDF2_SHA256':

scrypt-nosse.o(.text+0x8c8): In function `smix':
/arpa/tz/t/tycho/src/scrypt-1.1.1/scrypt-nosse.c:220: undefined reference to `le32enc'

cperciva · 7 months ago

If you edit config.h and take out the HAVE_SYS_ENDIAN_H line, does that unbreak the build?

royce · 7 months ago

Yes, it now builds. Now getting the same error as da44en's Solaris 10 test:

scrypt: Error reading clocks: Invalid argument

royce · 7 months ago

Compiles and verifies fine on CentOS 4.7 i686:

$ uname -a
Linux vps.alpca.org 2.6.18-028stab060.8 #1 SMP Mon Feb 9 20:25:36 MSK 2009 i686 i686 i386 GNU/Linux
$ cat /proc/version
Linux version 2.6.18-028stab060.8 (root@rhel5-64-build) (gcc version 4.1.2 20070626 (Red Hat 4.1.2-14)) #1 SMP Mon Feb 9 20:25:36 MSK 2009
$ cat /etc/redhat-release
CentOS release 4.7 (Final)

s0xxx · 7 months ago

nbsd$ uname -rms
NetBSD 5.0_STABLE i386

nbsd$ ./configure
checking for a BSD-compatible install... /usr/bin/install -c
checking whether build environment is sane... yes
checking for a thread-safe mkdir -p... config.aux/install-sh -c -d
checking for gawk... no
checking for mawk... no
checking for nawk... no
checking for awk... awk
checking whether make sets $(MAKE)... yes
checking whether to enable maintainer-specific portions of Makefiles... no
checking for gcc... gcc
checking for C compiler default output file name... a.out
checking whether the C compiler works... yes
checking whether we are cross compiling... no
checking for suffix of executables...
checking for suffix of object files... o
checking whether we are using the GNU C compiler... yes
checking whether gcc accepts -g... yes
checking for gcc option to accept ISO C89... none needed
checking for style of include used by make... GNU
checking dependency style of gcc... gcc3
checking how to run the C preprocessor... gcc -E
checking for grep that handles long lines and -e... /usr/bin/grep
checking for egrep... /usr/bin/grep -E
checking for ANSI C header files... yes
checking for sys/types.h... yes
checking for sys/stat.h... yes
checking for stdlib.h... yes
checking for string.h... yes
checking for memory.h... yes
checking for strings.h... yes
checking for inttypes.h... yes
checking for stdint.h... yes
checking for unistd.h... yes
checking err.h usability... yes
checking err.h presence... yes
checking for err.h... yes
checking sys/endian.h usability... yes
checking sys/endian.h presence... yes
checking for sys/endian.h... yes
checking sys/sysinfo.h usability... no
checking sys/sysinfo.h presence... no
checking for sys/sysinfo.h... no
checking for clock_gettime in -lrt... yes
checking for clock_gettime... yes
checking for sysinfo... no
checking for posix_memalign... yes
checking for struct sysinfo... no
checking for struct sysinfo.mem_unit... no
configure: creating ./config.status
config.status: creating Makefile
config.status: creating config.h
config.status: executing depfiles commands


nbsd$ make
make all-am
gcc -DHAVE_CONFIG_H -I. -g -O2 -MT main.o -MD -MP -MF .deps/main.Tpo -c -o main.o main.c
mv -f .deps/main.Tpo .deps/main.Po
gcc -DHAVE_CONFIG_H -I. -g -O2 -MT scryptenc.o -MD -MP -MF .deps/scryptenc.Tpo -c -o scryptenc.o scryptenc.c
mv -f .deps/scryptenc.Tpo .deps/scryptenc.Po
gcc -DHAVE_CONFIG_H -I. -g -O2 -MT crypto_aesctr.o -MD -MP -MF .deps/crypto_aesctr.Tpo -c -o crypto_aesctr.o crypto_aesctr.c
mv -f .deps/crypto_aesctr.Tpo .deps/crypto_aesctr.Po
gcc -DHAVE_CONFIG_H -I. -g -O2 -MT sha256.o -MD -MP -MF .deps/sha256.Tpo -c -o sha256.o sha256.c
mv -f .deps/sha256.Tpo .deps/sha256.Po
gcc -DHAVE_CONFIG_H -I. -g -O2 -MT memlimit.o -MD -MP -MF .deps/memlimit.Tpo -c -o memlimit.o memlimit.c
mv -f .deps/memlimit.Tpo .deps/memlimit.Po
gcc -DHAVE_CONFIG_H -I. -g -O2 -MT warn.o -MD -MP -MF .deps/warn.Tpo -c -o warn.o warn.c
mv -f .deps/warn.Tpo .deps/warn.Po
gcc -DHAVE_CONFIG_H -I. -g -O2 -MT scrypt_cpuperf.o -MD -MP -MF .deps/scrypt_cpuperf.Tpo -c -o scrypt_cpuperf.o scrypt_cpuperf.c
mv -f .deps/scrypt_cpuperf.Tpo .deps/scrypt_cpuperf.Po
gcc -DHAVE_CONFIG_H -I. -g -O2 -MT scrypt-nosse.o -MD -MP -MF .deps/scrypt-nosse.Tpo -c -o scrypt-nosse.o scrypt-nosse.c
mv -f .deps/scrypt-nosse.Tpo .deps/scrypt-nosse.Po
gcc -g -O2 -o scrypt main.o scryptenc.o crypto_aesctr.o sha256.o memlimit.o warn.o scrypt_cpuperf.o scrypt-nosse.o -lcrypto -lrt


nbsd$ ./scrypt enc ./Toaster_NetBSD.jpg ./Toaster_NetBSD.jpg.enc
Please enter passphrase:
Please confirm passphrase:
scrypt: Error reading clocks: Invalid argument

cperciva · 7 months ago

Thanks! This looks like a rather difficult bug to figure out via the commenting system here -- could you send an email to me (cperciva@tarsnap.com)?

Marshall Pierce · 7 months ago

Gentoo linux 2.6.27, gcc 4.3.2, glibc 2.8. Works succesfully.

akira · 7 months ago

% set OS
OS=Windows_NT

% scrypt.exe
usage: scrypt {enc | dec} [...] infile [outfile]

i have sent colin the patch, lets see what goes into scrypt :)

Kenji Rikitake · 7 months ago

Ubuntu 9.04, gcc 4.3.3. The scrypt was built and worked OK.

uname -a
Linux bluemax-ubuntu 2.6.28-12-server #43-Ubuntu SMP Fri May 1 20:28:32 UTC 2009 i686 GNU/Linux
Intel(R) Core2 DuO CPU E6300 @ 1.86GHz stepping 06

royce · 7 months ago

$ uname -a
CYGWIN_NT-5.1 mycroft 1.5.25(0.156/4/2) 2008-06-12 19:34 i686 Cygwin

$ ./scrypt enc scrypt.h scrypt.h.enc
Please enter passphrase: Please confirm passphrase: scrypt: Error reading salt: No error

I threw some fprintfs into getsalt(), and neither successes nor failures seem to be triggered, which makes me think that something other than getsalt() is returning 4 to scryptenc_file. Take that with a giant grain of salt, though; I know about two weeks' worth of C. :-)

cperciva · 7 months ago

Does /dev/urandom exist on your Cygwin machine?

royce · 7 months ago

Yep - I should have said as much.

$ cat /dev/urandom | strings | head -1
h>E73

akira · 7 months ago

i will look into that issue asap. i had to rewrite several parts of the code to work on native win32/64, /dev/urandom was one "issue".

colin, have my mails reached your spam folder?

cperciva · 7 months ago

Yes, I received your emails -- sorry for not replying. I wanted to get the "easy" unix portability issues out of the way before I tackled the Windows issues.

Tim · 7 months ago

Solaris 9: SunOS myrtle 5.9 Generic_122300-35 sun4u sparc SUNW,Sun-Fire-480R
(whether you care about older versions of Solaris like this one is up to you :-)

make[1]: Entering directory `/tmp/scrypt-1.1.2'
gcc -DHAVE_CONFIG_H -I. -g -O2 -MT main.o -MD -MP -MF .deps/main.Tpo -c -o main.o main.c
main.c:27:20: error: stdint.h: No such file or directory
make[1]: *** [main.o] Error 1
make[1]: Leaving directory `/tmp/scrypt-1.1.2'
make: *** [all] Error 2

cperciva · 7 months ago

Yes, I'm aware of this problem, and at the moment I don't think I want to fix it. Solaris 9 is pretty old, and stdint.h is a required C99 header, so odds are that if Solaris 9 is missing that it will also miss lots of other important bits.

That said, it's possible that if you install a newer (C99) compiler, it will bring a stdint.h implementation with it; so you might want to try that.