On Tue, Mar 20, 2018 at 11:53:08PM +0200, Arnon Warshavsky wrote:
> The purpose of this patch is to cleanup the library code
> from paths that end up aborting the process,
> and move to checking error values, in order to allow the running process
> perform an orderly tear-down or other mitigation of the event.
>
> This patch modifies the majority of rte_panic calls under lib and drivers,
> and replaces them with a new variation of rte_panic macro
> that does not abort and returns an error value
> that can be propagated up the call stack.
>
> - Focus was given to the dpdk initialization path
> - Some of the panic calls within drivers were left in place where
> the call is from within an interrupt or calls that are on the data path,
> where there is no simple applicative route to propagate
> the error to termination.
> These should be handled by the driver maintainers.
> I would like to define a device health state that can be monitored from
> the side,and this will be an independent patch.
> - No change took place in example and test files
> - No change took place for debug assertions calling panic
> - Some previously void functions calling panic where changed
> to return a value,with callers modified accordingly.
>
> An additional independent patch to devtools/checkpatches.sh
> will be submitted in order to prevent new additions of calls to rte_panic
> under lib and drivers.
>
> Keep calm and don't panic.
>
> ---
> V2: Fix typos in commit message
> ---
> V3: More commit message formatting
>
> Signed-off-by: Arnon Warshavsky <ar...@qwilt.com>
> ---
> drivers/crypto/dpaa2_sec/dpaa2_sec_dpseci.c | 2 +-
> drivers/crypto/dpaa_sec/dpaa_sec.c | 2 +-
> drivers/net/bonding/rte_eth_bond_8023ad.c | 23 +++++---
> drivers/net/bonding/rte_eth_bond_8023ad_private.h | 2 +-
> drivers/net/bonding/rte_eth_bond_api.c | 20 +++++--
> drivers/net/bonding/rte_eth_bond_pmd.c | 5 +-
> drivers/net/bonding/rte_eth_bond_private.h | 2 +-
> drivers/net/e1000/e1000_ethdev.h | 2 +-
> drivers/net/e1000/igb_ethdev.c | 3 +-
> drivers/net/e1000/igb_pf.c | 10 ++--
> drivers/net/ixgbe/ixgbe_ethdev.c | 3 +-
> drivers/net/ixgbe/ixgbe_ethdev.h | 2 +-
> drivers/net/ixgbe/ixgbe_pf.c | 6 +-
> lib/librte_eal/bsdapp/eal/eal.c | 54 ++++++++++-------
> lib/librte_eal/bsdapp/eal/eal_debug.c | 14 +++++
> lib/librte_eal/bsdapp/eal/eal_thread.c | 43 +++++++++-----
> lib/librte_eal/common/eal_common_memzone.c | 2 +-
> lib/librte_eal/common/eal_thread.h | 4 +-
> lib/librte_eal/common/include/rte_debug.h | 20 +++++++
> lib/librte_eal/common/rte_malloc.c | 6 +-
> lib/librte_eal/linuxapp/eal/eal.c | 72
> ++++++++++++++---------
> lib/librte_eal/linuxapp/eal/eal_debug.c | 14 +++++
> lib/librte_eal/linuxapp/eal/eal_hugepage_info.c | 19 ++++--
> lib/librte_eal/linuxapp/eal/eal_interrupts.c | 16 +++--
> lib/librte_eal/linuxapp/eal/eal_thread.c | 43 +++++++++-----
> lib/librte_eal/rte_eal_version.map | 1 +
> lib/librte_ether/rte_ethdev.c | 10 +++-
> lib/librte_eventdev/rte_eventdev_pmd_pci.h | 2 +-
> lib/librte_eventdev/rte_eventdev_pmd_vdev.h | 6 +-
> lib/librte_kni/rte_kni.c | 18 ++++--
> lib/librte_kni/rte_kni_fifo.h | 6 +-
> 31 files changed, 291 insertions(+), 141 deletions(-)
>
> diff --git a/drivers/crypto/dpaa2_sec/dpaa2_sec_dpseci.c
> b/drivers/crypto/dpaa2_sec/dpaa2_sec_dpseci.c
> index 9a790dd..9fda47c 100644
> --- a/drivers/crypto/dpaa2_sec/dpaa2_sec_dpseci.c
> +++ b/drivers/crypto/dpaa2_sec/dpaa2_sec_dpseci.c
> @@ -2865,7 +2865,7 @@ struct rte_security_ops dpaa2_sec_security_ops = {
> rte_socket_id());
>
> if (cryptodev->data->dev_private == NULL)
> - rte_panic("Cannot allocate memzone for private "
> + rte_panic_return("Cannot allocate memzone for private "
> "device data");
> }
To echo Thomas' previous comment, I'm not sure I like the addition of this
new "panic_return" macro. It seems an artificial hack just to have a 1:1
replacement for rte_panic. I think I'd rather an explicit log and then
return call.
/Bruce
