[Pkg-gnupg-commit] [gnupg2] 79/124: scd: Fix timeout handling for key generation.

[Daniel Kahn Gillmor]

This is an automated email from the git hooks/post-receive script.

dkg pushed a commit to branch experimental
in repository gnupg2.

commit 0848cfcce738150b53bfb65b78efc1e6dc9f3d26
Author: NIIBE Yutaka <gniibe at fsij.org>
Date:   Mon Mar 27 11:25:00 2017 +0900

    scd: Fix timeout handling for key generation.
    
    * scd/ccid-driver.c (CCID_CMD_TIMEOUT): Back to original value.
    (CCID_CMD_TIMEOUT_LONGER): New.
    (ccid_transceive): Add kludge for key generation.
    
    --
    
    At key generation of longer key size, OpenPGP card sends back "time
    extension" with BWI=100, which is unusual value in the protocol and it
    actually requires host having longer timeout like 43 seconds.
    
    Reported-by: Szczepan Zalega <szczepan at nitrokey.com>
    Fixes-commit: 6510df3a7cd2b5bf44fac1e4d50ee54b8c897daa
    Signed-off-by: NIIBE Yutaka <gniibe at fsij.org>
---
 scd/ccid-driver.c | 18 +++++++++++++-----
 1 file changed, 13 insertions(+), 5 deletions(-)

diff --git a/scd/ccid-driver.c b/scd/ccid-driver.c
index 40b2599..a471adf 100644
--- a/scd/ccid-driver.c
+++ b/scd/ccid-driver.c
@@ -108,8 +108,10 @@
 */
 #define CCID_MAX_BUF (2048+7+10)
 
-/* CCID command timeout.  OpenPGPcard v2.1 requires timeout of 13 seconds.  */
-#define CCID_CMD_TIMEOUT (13*1000)
+/* CCID command timeout.  */
+#define CCID_CMD_TIMEOUT (5*1000)
+/* OpenPGPcard v2.1 requires huge timeout for key generation.  */
+#define CCID_CMD_TIMEOUT_LONGER (60*1000)
 
 /* Depending on how this source is used we either define our error
    output to go to stderr or to the GnuPG based logging functions.  We
@@ -3394,6 +3396,7 @@ ccid_transceive (ccid_driver_t handle,
   int retries = 0;
   int resyncing = 0;
   int nad_byte;
+  int wait_more = 0;
 
   if (!nresp)
     nresp = &dummy_nresp;
@@ -3407,7 +3410,7 @@ ccid_transceive (ccid_driver_t handle,
          but the Windows driver does it this way.  Tested using a
          CM6121.  This method works also for the Cherry XX44
          keyboards; however there are problems with the
-         ccid_tranceive_secure which leads to a loss of sync on the
+         ccid_transceive_secure which leads to a loss of sync on the
          CCID level.  If Cherry wants to make their keyboard work
          again, they should hand over some docs. */
       if ((handle->id_vendor == VENDOR_OMNIKEY
@@ -3503,8 +3506,8 @@ ccid_transceive (ccid_driver_t handle,
 
       msg = recv_buffer;
       rc = bulk_in (handle, msg, sizeof recv_buffer, &msglen,
-                    via_escape? RDR_to_PC_Escape : RDR_to_PC_DataBlock,
-                    seqno, CCID_CMD_TIMEOUT, 0);
+                    via_escape? RDR_to_PC_Escape : RDR_to_PC_DataBlock, seqno,
+                    wait_more? CCID_CMD_TIMEOUT_LONGER: CCID_CMD_TIMEOUT, 0);
       if (rc)
         return rc;
 
@@ -3678,6 +3681,11 @@ ccid_transceive (ccid_driver_t handle,
             {
               /* Wait time extension request. */
               unsigned char bwi = tpdu[3];
+
+              /* Check if it's unsual value which can't be expressed in ATR.  */
+              if (bwi > 15)
+                wait_more = 1;
+
               msg = send_buffer;
               tpdu = msg + hdrlen;
               tpdu[0] = nad_byte;

-- 
Alioth's /usr/local/bin/git-commit-notice on /srv/git.debian.org/git/pkg-gnupg/gnupg2.git