Drivers with the flag RTE_PCI_DRV_MULTIPLE enabled will be bound several
times to the same device until they return an error.
These drivers must internally keep track of each device state in order for
this to work reliably.
This flag currently required for Ethernet adapters with several physical
ports but a single address on the PCI bus.
Signed-off-by: Adrien Mazarguil <adrien.mazarguil@6wind.com>
Acked-by: Ivan Boule <ivan.boule@6wind.com>
Acked-by: Damien Millescamps <damien.millescamps@6wind.com>
The returned value of lseek is a virtual address
which can be different from the offset.
Indeed, if the return address has a 64-bit canonical form,
the 16 higher bits are all 1 if bit 47 is 1.
So the check was wrong. It is better to test against an error value.
Signed-off-by: Thomas Monjalon <thomas.monjalon@6wind.com>
Acked-by: Damien Millescamps <damien.millescamps@6wind.com>
In rte.toolchain-compat.mk, ensure MACHINE_CFLAGS is initialized when
using GCC 4.7. Update the GCC 4.6 test to also include GCC 4.7.
Signed-off-by: Krzysztof Witek <krzysztof.witek@6wind.com>
Signed-off-by: Zijie Pan <zijie.pan@6wind.com>
Acked-by: Ivan Boule <ivan.boule@6wind.com>
Acked-by: Adrien Mazarguil <adrien.mazarguil@6wind.com>
Using gcc is wrong, we should use the cross gcc to detect the
version of gcc.
Signed-off-by: Samuel Gauthier <samuel.gauthier@6wind.com>
Acked-by: Thomas Monjalon <thomas.monjalon@6wind.com>
test-pmd txonly leaks mbuf from the pool.
The function tx_mbuf_alloc() does not change the refcnt
and the refcnt is 0 when it is first allocated.
However, rte_pktmbuf_free_seg called by the driver's xmit code decrements
reference count to -1. So mbuf never goes back to the pool.
As a result, txonly can't send packets after it exhausts the mempool.
The function tx_mbuf_alloc() was getting mbuf directly from mempool and so
was bypassing mbuf API.
The dedicated function is rte_pktmbuf_alloc() but it is much slower because
it does unnecessary initializations in rte_pktmbuf_reset().
By using the internal API __rte_mbuf_raw_alloc(), refcnt is correctly handled
without adding too much overload.
Signed-off-by: Dongsu Han <dongsuh@cs.cmu.edu>
Signed-off-by: Thomas Monjalon <thomas.monjalon@6wind.com>
Acked-by: Adrien Mazarguil <adrien.mazarguil@6wind.com>
