Since I experienced a similar issue when using curl with sftp I run the
sample with some debug and got to the point where in libssh2_sftp.h the
maximal packet length is defined as 4. Where does this maximum come from
? Is it an arbitrary value or somewhere defined in the ssh/sftp protoco
libssh2_htonu64(s, handle->u.file.offset);
s += 8;
libssh2_htonu32(s, buffer_maxlen);
s += 4;
sftp->read_state = libssh2_NB_state_created;
}
"Markus Moeller" wrote in
message news:gi0gi8$lk...@ger.gmane.org...
> Since I e
npgbe.fr...
> On Sat, 13 Dec 2008, Markus Moeller wrote:
>
> (Please trim your quotes)
>
>> The server can only answer with a size which was set before by the client
>> (otherwise it is a server bug). As far as I understand the buffer size
>> is
>> send as pa
[libssh2] SFTP Subsystem: Requesting 16384 bytes from SFTP handle
[libssh2] SFTP Subsystem: Asking for 103 packet
[libssh2] SFTP Subsystem: Asking for 101 packet
[libssh2] SFTP Subsystem: Waiting for packet
"Markus Moeller" wrote in
message news:gi12or$9q...@ger.gmane.org...
> Th
Should I open a bug for this on sourceforge ?
Markus
"Markus Moeller" wrote in
message news:gi12or$9q...@ger.gmane.org...
> The test from Alexander works. I need to check on the curl issue again.
> Unfortunately the server I had the issue with isn't available at the
>
"Daniel Stenberg" wrote in message
news:alpine.lrh.2.00.0812182112120.22...@yvahk3.pbagnpgbe.fr...
> On Thu, 18 Dec 2008, Markus Moeller wrote:
>
>> Should I open a bug for this on sourceforge ?
>
> I don't think submitting anything there helps a lot - a
"Daniel Stenberg" wrote in message
news:alpine.deb.1.10.0902040905110.27...@yvahk2.pbagnpgbe.fr...
> On Tue, 3 Feb 2009, Markus Moeller wrote:
>
>>> [libssh2] Connection: Sending 39 bytes on channel 0/0, stream_id=0
>>> [libssh2] Connection: libssh2_packet_w
I did some more debug. Here are my questions:
Normal:
Curl_write calls Curl_scp_send
MMM Curl_scp_send after libssh2_channel_write len/nwrite 16384/16384
MMM Curl_write after Curl_scp_send len/nwrite 16384/16384
MMM Curl_scp_send after libssh2_channel_write len/nwrite 16384/16384
MMM
"Markus Moeller" wrote in
message news:gmi9el$hi...@ger.gmane.org...
>I did some more debug. Here are my questions:
>
> Normal:
> Curl_write calls Curl_scp_send
>
> MMM Curl_scp_send after libssh2_channel_write len/nwrite 16384/16384
> MMM Curl_write
I think I now know under which rare circumstances this happens and it is a
bug in libssh2.
I can simulate it with (e.g. a packet is not completly written and the
attempt to write the rest creates EAGAIN):
In below simulation the 10 packet is reduced by 30 bytes and the attempt to
write the 30
10 matches
Mail list logo
