2017-09-26, 16:56
@Milhouse
In the samba 4.7.0 PR I don't think that this is correct:
I may be very mistaken tho.
From looking at the test code itself it looks like the message is misleading, it doesn't look for errcode value itself, but for returncode which will be 0 (have EILSEQ) or -1 (doesn't have EILSEQ)
The define resultiing from this is
Which is the same value I get in native compilation on x64 box.
Compiling the test code natively on the pi with glibc 2.26, no external libiconv I get retcode 0 and I think that the retcode value should be put in cache.
Cross-compiling samba with value 0 in cache succeds, however I haven't had time to thoroughly test it.
Again I may be mistaken.
In the samba 4.7.0 PR I don't think that this is correct:
Code:
Checking errno of iconv for illegal multibyte sequence: "84"
I may be very mistaken tho.
From looking at the test code itself it looks like the message is misleading, it doesn't look for errcode value itself, but for returncode which will be 0 (have EILSEQ) or -1 (doesn't have EILSEQ)
The define resultiing from this is
Code:
#define HAVE_ICONV_ERRNO_ILLEGAL_MULTIBYTE 1
Compiling the test code natively on the pi with glibc 2.26, no external libiconv I get retcode 0 and I think that the retcode value should be put in cache.
Cross-compiling samba with value 0 in cache succeds, however I haven't had time to thoroughly test it.
Again I may be mistaken.