On Fri, Dec 04, 2015 at 09:18:38AM +0100, Bernd Helm wrote: > On 12/03/2015 06:53 PM, Willy Tarreau wrote: > >Maybe we're facing a bug where the buffer wraps at the end or something > >like this. Bernd, if you still have the core, could you please issue > >"print *b" while in buffer_insert_line2() ? > yes, i still have the core. > > (gdb) frame 1 > #1 0x0000000000413349 in buffer_insert_line2 (b=0x1e47a70, > pos=0x1e47acb "ntent-Type: text/html; > charset=UTF-8\r\nTransfer-Encoding: chunked\r\n\r\n1b9\r\n<!DOCTYPE > html PUBLIC \"-//W3C//DTD XHTML 1.0 Transitional//EN\" > \"http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd\">\n<html"..., > str=0x513979 "Connection: close", > len=17) at src/buffer.c:126 > (gdb) print *b > $1 = {p = 0x0, size = 822083584, i = 64, o = 2818572288, data = 0x1e47a84 > "z\344\001"} > > if you need more information, let me know. thank you.
The buffer was corrupted :-( I guess it was corrupted just before doing this buffer_insert_line2(), causing it to crash. Buffers are manipulated a lot, so it must have been still good a few nanoseconds before calling this. I'll review the code in this area in case I can find any hint relevant to your config. Thanks, Willy