<rdar://problem/15296388>
Fix a crasher that would occur if one tried to read memory as characters of some size != 1, e.g.
x -f c -s 10 buffer
This commit tries to do the right thing and uses the byte-size as the number of elements, unless both are specified and the number of elements is != 1
In this latter case (e.g. x -f c -s 10 -c 3 buffer) one could multiply the two and read 30 characters, but it seems a stretch in mind reading.
llvm-svn: 193659
diff --git a/lldb/source/Commands/CommandObjectMemory.cpp b/lldb/source/Commands/CommandObjectMemory.cpp
index d7e59fb..e804b4a 100644
--- a/lldb/source/Commands/CommandObjectMemory.cpp
+++ b/lldb/source/Commands/CommandObjectMemory.cpp
@@ -860,16 +860,28 @@
Format format = m_format_options.GetFormat();
if ( ( (format == eFormatChar) || (format == eFormatCharPrintable) )
- && (item_byte_size != 1)
- && (item_count == 1))
+ && (item_byte_size != 1))
{
- // this turns requests such as
- // memory read -fc -s10 -c1 *charPtrPtr
- // which make no sense (what is a char of size 10?)
- // into a request for fetching 10 chars of size 1 from the same memory location
- format = eFormatCharArray;
- item_count = item_byte_size;
- item_byte_size = 1;
+ // if a count was not passed, or it is 1
+ if (m_format_options.GetCountValue().OptionWasSet() == false || item_count == 1)
+ {
+ // this turns requests such as
+ // memory read -fc -s10 -c1 *charPtrPtr
+ // which make no sense (what is a char of size 10?)
+ // into a request for fetching 10 chars of size 1 from the same memory location
+ format = eFormatCharArray;
+ item_count = item_byte_size;
+ item_byte_size = 1;
+ }
+ else
+ {
+ // here we passed a count, and it was not 1
+ // so we have a byte_size and a count
+ // we could well multiply those, but instead let's just fail
+ result.AppendErrorWithFormat("reading memory as characters of size %zu is not supported", item_byte_size);
+ result.SetStatus(eReturnStatusFailed);
+ return false;
+ }
}
assert (output_stream);